[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-open-edge-platform--dlstreamer":3,"tool-open-edge-platform--dlstreamer":62},[4,18,26,36,46,54],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",160015,2,"2026-04-18T11:30:52",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":42,"last_commit_at":43,"category_tags":44,"status":17},8272,"opencode","anomalyco\u002Fopencode","OpenCode 是一款开源的 AI 编程助手（Coding Agent），旨在像一位智能搭档一样融入您的开发流程。它不仅仅是一个代码补全插件，而是一个能够理解项目上下文、自主规划任务并执行复杂编码操作的智能体。无论是生成全新功能、重构现有代码，还是排查难以定位的 Bug，OpenCode 都能通过自然语言交互高效完成，显著减少开发者在重复性劳动和上下文切换上的时间消耗。\n\n这款工具专为软件开发者、工程师及技术研究人员设计，特别适合希望利用大模型能力来提升编码效率、加速原型开发或处理遗留代码维护的专业人群。其核心亮点在于完全开源的架构，这意味着用户可以审查代码逻辑、自定义行为策略，甚至私有化部署以保障数据安全，彻底打破了传统闭源 AI 助手的“黑盒”限制。\n\n在技术体验上，OpenCode 提供了灵活的终端界面（Terminal UI）和正在测试中的桌面应用程序，支持 macOS、Windows 及 Linux 全平台。它兼容多种包管理工具，安装便捷，并能无缝集成到现有的开发环境中。无论您是追求极致控制权的资深极客，还是渴望提升产出的独立开发者，OpenCode 都提供了一个透明、可信",144296,1,"2026-04-16T14:50:03",[13,45],"插件",{"id":47,"name":48,"github_repo":49,"description_zh":50,"stars":51,"difficulty_score":32,"last_commit_at":52,"category_tags":53,"status":17},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",109154,"2026-04-18T11:18:24",[14,15,13],{"id":55,"name":56,"github_repo":57,"description_zh":58,"stars":59,"difficulty_score":32,"last_commit_at":60,"category_tags":61,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[45,13,15,14],{"id":63,"github_repo":64,"name":65,"description_en":66,"description_zh":67,"ai_summary_zh":67,"readme_en":68,"readme_zh":69,"quickstart_zh":70,"use_case_zh":71,"hero_image_url":72,"owner_login":73,"owner_name":74,"owner_avatar_url":75,"owner_bio":76,"owner_company":77,"owner_location":77,"owner_email":77,"owner_twitter":77,"owner_website":77,"owner_url":78,"languages":79,"stars":110,"forks":111,"last_commit_at":112,"license":113,"difficulty_score":10,"env_os":114,"env_gpu":115,"env_ram":116,"env_deps":117,"category_tags":126,"github_topics":127,"view_count":32,"oss_zip_url":77,"oss_zip_packed_at":77,"status":17,"created_at":133,"updated_at":134,"faqs":135,"releases":164},9151,"open-edge-platform\u002Fdlstreamer","dlstreamer","Deep Learning Streamer (DL Streamer) Pipeline Framework is an open-source streaming media analytics framework, based on GStreamer* multimedia framework, for creating complex media analytics pipelines for the Cloud or at the Edge.","DL Streamer 是一款基于 GStreamer 构建的开源流媒体分析框架，专为在云端或边缘端打造复杂的媒体分析流水线而设计。它核心解决的是如何高效处理音视频流并从中提取价值的问题，能够实时检测、分类、追踪和统计画面中的人、物体及事件，广泛适用于零售分析、仓储管理、工业质检及安全监控等场景。\n\n这款工具主要面向需要开发视频智能分析应用的开发者与系统架构师。其独特亮点在于深度的硬件优化与灵活的扩展性：推理插件基于 OpenVINO™ 引擎，充分释放 Intel CPU、GPU 及 VPU 的性能潜力；视频编解码利用 VA-API 实现 GPU 加速；图像处理则融合了 OpenCV 与 DPC++ 技术。此外，DL Streamer 支持 OpenVINO IR 和 ONNX 多种模型格式，兼容 YOLO、ResNet 等主流算法骨架，并提供丰富的 C\u002FC++ 与 Python 示例及预训练模型资源，帮助用户快速搭建从数据采集到智能决策的全链路应用，大幅降低高性能媒体分析系统的开发门槛。","# Intel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework\n## DL Streamer is now part of Open Edge Platform\n\n### Overview\n\u003Cdiv align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-edge-platform_dlstreamer_readme_3c87e901d501.gif\" width=900\u002F>\u003C\u002Fdiv>\n\n[Deep Learning Streamer](.\u002Fdocs\u002Fuser-guide\u002Findex.md) (**DL Streamer**) Pipeline Framework is an open-source streaming media analytics framework, based on [GStreamer*](https:\u002F\u002Fgstreamer.freedesktop.org) multimedia framework, for creating complex media analytics pipelines for the Cloud or at the Edge.\n\n**Media analytics** is the analysis of audio & video streams to detect, classify, track, identify and count objects, events and people. The analyzed results can be used to take actions, coordinate events, identify patterns and gain insights across multiple domains: retail store and events facilities analytics, warehouse and parking management, industrial inspection, safety and regulatory compliance, security monitoring, and many other.\n\n## Backend libraries\nDL Streamer Pipeline Framework is optimized for performance and functional interoperability between GStreamer* plugins built on various backend libraries\n* Inference plugins use [OpenVINO™ inference engine](https:\u002F\u002Fdocs.openvino.ai) optimized for Intel CPU, GPU and VPU platforms\n* Video decode and encode plugins utilize [GPU-acceleration based on VA-API](https:\u002F\u002Fgithub.com\u002FGStreamer\u002Fgstreamer-vaapi)\n* Image processing plugins based on [OpenCV](https:\u002F\u002Fopencv.org\u002F) and [DPC++](https:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fdevelop\u002Fdocumentation\u002Foneapi-programming-guide\u002Ftop\u002Foneapi-programming-model\u002Fdata-parallel-c-dpc.html)\n* Hundreds other [GStreamer* plugins](https:\u002F\u002Fgstreamer.freedesktop.org\u002Fdocumentation\u002Fplugins_doc.html) built on various open-source libraries for media input and output, muxing and demuxing, decode and encode\n\n[This page](.\u002Fdocs\u002Fuser-guide\u002Felements\u002Felements.md) contains a list of elements provided in this repository.\n\n## Prerequisites\nPlease refer to [System Requirements](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Fsystem_requirements.md) for details.\n\n## Installation\nPlease refer to [Install Guide](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Finstall\u002Finstall_guide_ubuntu.md) for installation options\n1. [Install APT packages](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Finstall\u002Finstall_guide_ubuntu.md#option-1-install-intel-dl-streamer-pipeline-framework-from-debian-packages-using-apt-repository)\n2. [Run Docker image](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Finstall\u002Finstall_guide_ubuntu.md#option-2-install-docker-image-from-docker-hub-and-run-it)\n3. [Compile from source code](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fadvanced_install\u002Fadvanced_install_guide_compilation.md)\n4. [Build Docker image from source code](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fadvanced_install\u002Fadvanced_build_docker_image.md)\n\nTo see the full list of installed components check the [dockerfile content for Ubuntu24](https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Fblob\u002Fmain\u002Fdocker\u002Fubuntu\u002Fubuntu24.Dockerfile)\n\n## Samples\n[Samples](https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Ftree\u002Fmain\u002Fsamples) available for C\u002FC++ and Python programming, and as gst-launch command lines and scripts.\n\n## Models\nDL Streamer supports models in OpenVINO™ IR and ONNX* formats, including VLMs, object detection, object classification, human pose detection, sound classification, semantic segmentation, and other use cases on SSD, MobileNet, YOLO, Tiny YOLO, EfficientDet, ResNet, FasterRCNN, and other backbones.\n\nSee the full list of [supported models](.\u002Fdocs\u002Fuser-guide\u002Fsupported_models.md), including models pre-trained with [Intel® Geti™ Software](\u003Chttps:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fdeveloper\u002Ftools\u002Ftiber\u002Fedge-platform\u002Fmodel-builder.html>), or explore over 70 pre-trained models in [OpenVINO™ Open Model Zoo](https:\u002F\u002Fdocs.openvino.ai\u002Flatest\u002Fomz_models_group_intel.html) with corresponding [model-proc files](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fdlstreamer\u002Ftree\u002Fmain\u002Fsamples\u002Fmodel_proc) (pre- and post-processing specifications).\n\n## Other Useful Links\n* [Get Started](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Fget_started_index.md)\n* [Developer Guide](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fdev_guide_index.md)\n* [API Reference](.\u002Fdocs\u002Fuser-guide\u002Fapi_ref\u002Fapi_reference.rst)\n\n---\n\\* Other names and brands may be claimed as the property of others.\n\n## License\n\nThe **DL Streamer** project is licensed under the [MIT License](.\u002FLICENSE) license.\n\n","# 英特尔® 深度学习流媒体（Intel® DL Streamer）管道框架\n## DL Streamer 现已成为 Open Edge Platform 的一部分\n\n### 概述\n\u003Cdiv align=\"center\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-edge-platform_dlstreamer_readme_3c87e901d501.gif\" width=900\u002F>\u003C\u002Fdiv>\n\n[深度学习流媒体](.\u002Fdocs\u002Fuser-guide\u002Findex.md)（**DL Streamer**）管道框架是一个开源的流媒体分析框架，基于 [GStreamer*](https:\u002F\u002Fgstreamer.freedesktop.org) 多媒体框架，用于在云端或边缘端构建复杂的媒体分析管道。\n\n**媒体分析**是对音频和视频流进行分析，以检测、分类、跟踪、识别和计数对象、事件和人员。分析结果可用于采取行动、协调事件、识别模式，并在多个领域中获得洞察：零售商店和活动场馆分析、仓库和停车场管理、工业检测、安全与法规遵从、安防监控等众多应用。\n\n## 后端库\nDL Streamer 管道框架针对性能以及基于不同后端库构建的 GStreamer* 插件之间的功能互操作性进行了优化：\n* 推理插件使用 [OpenVINO™ 推理引擎](https:\u002F\u002Fdocs.openvino.ai)，该引擎针对英特尔 CPU、GPU 和 VPU 平台进行了优化。\n* 视频解码和编码插件利用 [基于 VA-API 的 GPU 加速](https:\u002F\u002Fgithub.com\u002FGStreamer\u002Fgstreamer-vaapi)。\n* 图像处理插件基于 [OpenCV](https:\u002F\u002Fopencv.org\u002F) 和 [DPC++](https:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fdevelop\u002Fdocumentation\u002Foneapi-programming-guide\u002Ftop\u002Foneapi-programming-model\u002Fdata-parallel-c-dpc.html) 构建。\n* 此外，还有数百个其他 [GStreamer* 插件](https:\u002F\u002Fgstreamer.freedesktop.org\u002Fdocumentation\u002Fplugins_doc.html)，它们基于各种开源库，用于媒体的输入输出、多路复用与解复用、解码与编码。\n\n[此页面](.\u002Fdocs\u002Fuser-guide\u002Felements\u002Felements.md)包含本仓库提供的元素列表。\n\n## 先决条件\n详细信息请参阅 [系统要求](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Fsystem_requirements.md)。\n\n## 安装\n安装选项请参阅 [安装指南](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Finstall\u002Finstall_guide_ubuntu.md)：\n1. [安装 APT 包](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Finstall\u002Finstall_guide_ubuntu.md#option-1-install-intel-dl-streamer-pipeline-framework-from-debian-packages-using-apt-repository)\n2. [运行 Docker 镜像](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Finstall\u002Finstall_guide_ubuntu.md#option-2-install-docker-image-from-docker-hub-and-run-it)\n3. [从源代码编译](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fadvanced_install\u002Fadvanced_install_guide_compilation.md)\n4. [从源代码构建 Docker 镜像](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fadvanced_install\u002Fadvanced_build_docker_image.md)\n\n要查看已安装组件的完整列表，请参阅适用于 Ubuntu 24 的 [Dockerfile 内容](https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Fblob\u002Fmain\u002Fdocker\u002Fubuntu\u002Fubuntu24.Dockerfile)。\n\n## 示例\n[示例](https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Ftree\u002Fmain\u002Fsamples)提供 C\u002FC++ 和 Python 编程版本，以及 gst-launch 命令行和脚本形式。\n\n## 模型\nDL Streamer 支持 OpenVINO™ IR 和 ONNX* 格式的模型，涵盖视觉语言模型、目标检测、目标分类、人体姿态估计、声音分类、语义分割等多种应用场景，使用的骨干网络包括 SSD、MobileNet、YOLO、Tiny YOLO、EfficientDet、ResNet、FasterRCNN 等。\n\n完整的 [支持的模型列表](.\u002Fdocs\u002Fuser-guide\u002Fsupported_models.md)请参见，其中包括使用 [Intel® Geti™ 软件](\u003Chttps:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fdeveloper\u002Ftools\u002Ftiber\u002Fedge-platform\u002Fmodel-builder.html>)预训练的模型；您也可以在 [OpenVINO™ 开放模型动物园](https:\u002F\u002Fdocs.openvino.ai\u002Flatest\u002Fomz_models_group_intel.html)中探索超过 70 种预训练模型，并参考相应的 [model-proc 文件](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fdlstreamer\u002Ftree\u002Fmain\u002Fsamples\u002Fmodel_proc)（预处理和后处理规范）。\n\n## 其他实用链接\n* [快速入门](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Fget_started_index.md)\n* [开发者指南](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fdev_guide_index.md)\n* [API 参考](.\u002Fdocs\u002Fuser-guide\u002Fapi_ref\u002Fapi_reference.rst)\n\n---\n\\* 其他名称和品牌可能属于其各自的所有者。\n\n## 许可证\n\n**DL Streamer** 项目采用 [MIT 许可证](.\u002FLICENSE)授权。","# Intel® DL Streamer 快速上手指南\n\nIntel® Deep Learning Streamer (DL Streamer) 是一个基于 GStreamer 的开源流媒体分析框架，专为云端和边缘端的复杂媒体分析管道设计。它利用 OpenVINO™ 进行推理加速，支持对象检测、分类、跟踪等多种 AI 任务。\n\n## 1. 环境准备\n\n### 系统要求\n*   **操作系统**: 推荐 Ubuntu 20.04, 22.04 或 24.04 (64 位)。\n*   **硬件平台**: 兼容 Intel CPU、集成显卡 (iGPU)、独立显卡 (dGPU) 及 VPU (如 Movidius)。\n*   **驱动依赖**:\n    *   若使用 GPU 加速，需安装最新的 Intel GPU 驱动程序。\n    *   确保已安装 `libva-dev` 和 `gstreamer` 相关基础库。\n\n> **注意**: 详细硬件兼容性列表请参考官方 [System Requirements](.\u002Fdocs\u002Fuser-guide\u002Fget_started\u002Fsystem_requirements.md)。\n\n## 2. 安装步骤\n\nDL Streamer 提供多种安装方式，推荐初学者使用 **Docker** 方式以快速体验，生产环境可选择 **APT 包安装**。\n\n### 方式一：使用 Docker（推荐）\n\n这是最快捷的方式，无需配置本地依赖，直接拉取预构建镜像运行。\n\n```bash\n# 拉取最新稳定版镜像\ndocker pull ghcr.io\u002Fopen-edge-platform\u002Fdlstreamer:latest\n\n# 运行容器（挂载当前目录并启用 GPU 权限）\ndocker run --rm -it --device=\u002Fdev\u002Fdri:\u002Fdev\u002Fdri \\\n  -v $(pwd):\u002Fhome\u002Fuser\u002Fwork \\\n  ghcr.io\u002Fopen-edge-platform\u002Fdlstreamer:latest\n```\n\n### 方式二：通过 APT 包安装 (Ubuntu)\n\n适用于需要直接集成到本地系统的场景。\n\n```bash\n# 1. 添加 DL Streamer 仓库密钥\nsudo apt install -y wget gnupg\nwget -O - https:\u002F\u002Frepositories.intel.com\u002Fgpu\u002Fintel-graphics.key | \\\n  sudo gpg --dearmor --output \u002Fusr\u002Fshare\u002Fkeyrings\u002Fintel-graphics.gpg\n\n# 2. 添加仓库源 (以 Ubuntu 22.04 为例，请根据实际版本调整 codename)\necho \"deb [arch=amd64 signed-by=\u002Fusr\u002Fshare\u002Fkeyrings\u002Fintel-graphics.gpg] \\\n  https:\u002F\u002Frepositories.intel.com\u002Fgpu\u002Fubuntu jammy\u002Flts\u002F2350 unified\" | \\\n  sudo tee \u002Fetc\u002Fapt\u002Fsources.list.d\u002Fintel-gpu-jammy.list\n\n# 3. 更新包列表并安装\nsudo apt update\nsudo apt install -y intel-dlstreamer\n```\n\n### 方式三：源码编译\n\n如需自定义开发或修改底层代码，请参考 [编译指南](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fadvanced_install\u002Fadvanced_install_guide_compilation.md)。\n\n## 3. 基本使用\n\nDL Streamer 的核心是通过 GStreamer 管道串联视频输入、AI 推理和视频输出。以下是一个最简单的示例，使用预训练模型对视频流中的人脸进行检测并绘制边框。\n\n### 前置准备：下载模型\n确保已下载 OpenVINO 格式的人脸检测模型（例如 `face-detection-adas-0001`）。你可以从 [Open Model Zoo](https:\u002F\u002Fdocs.openvino.ai\u002Flatest\u002Fomz_models_group_intel.html) 获取，或使用 `omz_downloader` 工具：\n\n```bash\nomz_downloader --name face-detection-adas-0001\n```\n\n### 运行示例管道\n\n以下命令使用 `gst-launch-1.0` 构建一个完整的分析管道：\n1. 读取测试视频文件。\n2. 解码视频。\n3. 加载 OpenVINO 模型进行人脸检测。\n4. 将检测结果绘制在视频帧上。\n5. 显示输出窗口。\n\n```bash\ngst-launch-1.0 filesrc location=test_video.mp4 ! decodebin ! \\\ngvadetect model=models\u002Fintel\u002Fface-detection-adas-0001\u002FFP32\u002Fface-detection-adas-0001.xml \\\n            model-proc=models\u002Fintel\u002Fface-detection-adas-0001\u002Fface-detection-adas-0001.json \\\n            name=detect ! gvawatermark name=draw ! fpsdisplaysink video-sink=xvimagesink sync=false\n```\n\n**参数说明：**\n*   `gvadetect`: DL Streamer 的核心推理插件，用于加载模型并执行检测。\n*   `model`: 指向 `.xml` 格式的 OpenVINO 模型文件路径。\n*   `model-proc`: 指向预处理和后处理配置文件（通常与模型配套提供）。\n*   `gvawatermark`: 将推理结果（边界框、标签）绘制到视频帧上。\n*   `fpsdisplaysink`: 显示视频并实时打印 FPS 性能数据。\n\n### Python 开发示例\n除了命令行，DL Streamer 也支持 Python API 构建更复杂的逻辑。在 Docker 容器中，你可以参考 `samples\u002Fpython` 目录下的脚本直接运行：\n\n```bash\npython3 samples\u002Fpython\u002Fobject_detection.py --input test_video.mp4\n```\n\n---\n*更多高级用法、自定义元素开发及完整 API 参考，请访问 [Developer Guide](.\u002Fdocs\u002Fuser-guide\u002Fdev_guide\u002Fdev_guide_index.md)。*","某大型连锁超市需要在边缘端部署一套实时客流分析系统，以监控货架前的顾客停留时间并统计热区分布。\n\n### 没有 dlstreamer 时\n- **开发周期漫长**：团队需手动编写代码串联视频解码、AI 推理（如 YOLO 模型）和后处理逻辑，不同组件间接口适配耗时数周。\n- **硬件性能浪费**：难以充分利用 Intel CPU、GPU 或 VPU 的加速能力，导致多路视频流分析时帧率低下，无法满足实时性要求。\n- **运维部署复杂**：缺乏统一的流水线管理，每次更新算法模型或调整参数都需要重新编译整个应用，云端与边缘端环境一致性难保障。\n- **功能扩展困难**：若想增加“跌倒检测”或“声音分类”等新功能，往往需要重构底层架构，系统耦合度极高。\n\n### 使用 dlstreamer 后\n- **快速构建流水线**：基于 GStreamer 插件机制，通过简单的配置文件或命令行即可将解码、OpenVINO 推理和 OpenCV 后处理组装成复杂管道，开发时间缩短至几天。\n- **极致性能释放**：dlstreamer 自动调用 VA-API 进行硬编解码，并针对 Intel 架构优化推理引擎，轻松支持多路高清视频流的并发实时分析。\n- **灵活迭代部署**：支持热插拔式修改 pipeline 元素，更换模型或调整逻辑无需重新编译，配合 Docker 镜像可实现云边端一键同步部署。\n- **生态兼容性强**：直接复用数百个现有 GStreamer 插件及 Open Model Zoo 中的预训练模型，轻松扩展人脸重识别、语义分割等高级分析功能。\n\ndlstreamer 通过标准化的媒体分析流水线框架，让开发者从繁琐的底层集成中解放出来，专注于业务逻辑创新与边缘智能的高效落地。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fopen-edge-platform_dlstreamer_39fd45fe.png","open-edge-platform","Open Edge Platform","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fopen-edge-platform_bfc55d04.png","",null,"https:\u002F\u002Fgithub.com\u002Fopen-edge-platform",[80,84,88,92,96,100,103,106],{"name":81,"color":82,"percentage":83},"C++","#f34b7d",75.3,{"name":85,"color":86,"percentage":87},"C","#555555",9.8,{"name":89,"color":90,"percentage":91},"Python","#3572A5",6,{"name":93,"color":94,"percentage":95},"CMake","#DA3434",4.2,{"name":97,"color":98,"percentage":99},"Dockerfile","#384d54",1.7,{"name":101,"color":102,"percentage":99},"Shell","#89e051",{"name":104,"color":105,"percentage":42},"PowerShell","#012456",{"name":107,"color":108,"percentage":109},"Makefile","#427819",0.2,600,190,"2026-04-18T00:25:37","MIT","Linux (Ubuntu)","非必需但推荐用于加速。支持 Intel GPU\u002FVPU (通过 VA-API 和 OpenVINO)，未提及 NVIDIA\u002FCUDA 需求。","未说明",{"notes":118,"python":119,"dependencies":120},"该工具主要优化用于 Intel 硬件平台 (CPU, GPU, VPU)。视频编解码利用基于 VA-API 的 GPU 加速。支持 OpenVINO IR 和 ONNX 格式的模型。可通过 APT、Docker 或源码编译安装，官方文档重点提供了 Ubuntu 系统的安装指南。","支持 Python (具体版本未在 README 中明确，通常需 3.8+)",[121,122,123,124,125],"GStreamer","OpenVINO™ inference engine","VA-API (gstreamer-vaapi)","OpenCV","DPC++",[45,14],[128,129,130,131,132],"gstreamer","gstreamer-plugins","inference","intel","openvino","2026-03-27T02:49:30.150509","2026-04-19T03:05:03.357740",[136,141,145,150,154,159],{"id":137,"question_zh":138,"answer_zh":139,"source_url":140},41077,"运行 GStreamer 管道连接摄像头时出现 'Internal data stream error' 和 'not-negotiated' 错误怎么办？","该错误通常表示视频源（v4l2src）与解码器或后续元素之间的格式协商失败。请检查以下几点：\n1. 确认摄像头设备路径（如 \u002Fdev\u002Fvideo0）正确且未被其他程序占用。\n2. 尝试在 v4l2src 后显式添加 caps 过滤器指定格式，例如：`video\u002Fx-raw,format=NV12` 或 `video\u002Fx-raw,format=YUY2`。\n3. 确保安装了正确的 GStreamer 插件（如 gst-plugins-good, gst-plugins-bad）。\n4. 如果是 Intel RealSense 相机，可能需要特定的后端配置或固件更新。\n5. 检查环境变量 `XDG_RUNTIME_DIR` 是否已设置（虽然这通常导致警告而非致命错误，但在某些容器环境中可能影响管道初始化）。","https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Fissues\u002F410",{"id":142,"question_zh":143,"answer_zh":144,"source_url":140},41078,"如何在 DL Streamer 中正确使用自定义的 YOLOv8 模型？","使用自定义 YOLOv8 模型时，如果遇到模型过大或无法加载的错误，请注意：\n1. DL Streamer 对模型大小和输入分辨率有限制，确保模型已针对目标硬件（如 GPU\u002FCPU）进行优化（例如转换为 OpenVINO IR 格式并量化为 FP16 或 INT8）。\n2. 不要直接安装 ultralytics 包并在 Docker 内随意调用脚本，应遵循 DL Streamer 的模型准备流程。\n3. 使用官方提供的 `yolo_download.sh` 和 `yolo_detect.sh` 脚本作为参考，但需将模型路径指向你转换好的 OpenVINO IR 模型（.xml 和 .bin 文件）。\n4. 确保 `model-proc` JSON 配置文件正确匹配你的 YOLOv8 版本（输出层名称、锚点等），YOLOv8 的后处理逻辑可能与 v4\u002Fv5 不同，可能需要自定义 model-proc 文件。",{"id":146,"question_zh":147,"answer_zh":148,"source_url":149},41079,"在 GPU 上启用共享内存（vaapi-surface-sharing）时出现 DRM_IOCTL_I915_GEM_APERTURE 失败错误如何解决？","当在 GStreamer 管道中添加 `video\u002Fx-raw(memory:VASurface)` 和 `pre-process-backend=vaapi-surface-sharing` 参数时出现此错误，通常与 GPU 驱动或内核配置有关：\n1. 这是一个底层驱动问题，建议首先更新 Intel GPU 驱动程序和 Linux 内核到最新版本。\n2. 检查是否以特权模式运行 Docker 容器（需要 `--device \u002Fdev\u002Fdri` 和可能的 `--privileged` 标志）。\n3. 如果问题依旧，这可能是一个 OpenVINO 或 i915 驱动的 Bug。建议前往 OpenVINO GitHub 仓库提交 Issue，并提供复现步骤（包括使用的模型、视频文件、完整命令行参数）。\n4. 作为临时变通方案，可以尝试不使用共享内存后端，让数据在系统内存和显存之间拷贝，虽然性能会略有下降但能避免此崩溃。","https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Fissues\u002F366",{"id":151,"question_zh":152,"answer_zh":153,"source_url":149},41080,"为什么 YOLOv8 比 YOLOv4 Tiny 占用更多的 CPU 资源？如何优化性能？","YOLOv8 架构通常比 YOLOv4 Tiny 更复杂，导致计算量增加。若在多路摄像头（如 50 路）场景下 CPU 占用过高，可尝试以下优化：\n1. **模型量化**：将模型转换为 OpenVINO IR 格式并使用 FP16 或 INT8 量化，显著降低计算负载。\n2. **硬件加速**：确保 `device` 参数设置为 `GPU` 而不是 `CPU`，利用 Intel GPU 进行推理。\n3. **预处理卸载**：使用 `pre-process-backend=vaapi-surface-sharing` 将颜色空间转换和缩放操作卸载到 GPU，减少 CPU 负担。\n4. **调整输入分辨率**：降低输入视频的分辨率，较小的输入尺寸能大幅提升推理速度。\n5. **批处理**：如果应用场景允许，适当增加 batch size 以提高吞吐量（需权衡延迟）。\n6. 对比测试时，建议使用 `benchmark_app` 工具单独测试模型性能，排除 GStreamer 管道开销的影响。",{"id":155,"question_zh":156,"answer_zh":157,"source_url":158},41081,"使用自定义 YOLOv4 模型时，如何正确配置 model-proc JSON 文件以避免激活函数错误？","当使用自定义训练的 YOLOv4 模型（特别是从 Darknet 转换而来）时，默认的 `yolo-v4-tf.json` 配置可能不适用，导致输出处理错误。解决方法如下：\n1. **检查激活函数**：如果你的模型输出层已经包含了 Sigmoid 或 Softmax 激活，或者不需要这些激活，必须在 model-proc 文件中禁用它们。将 `\"do_cls_softmax\": true` 和 `\"output_sigmoid_activation\": true` 改为 `false` 或直接删除这两行。\n2. **调整网格数量**：根据模型输入分辨率修改 `cells_number` 参数（例如 416 输入对应 13x13 网格，则设为 13）。\n3. **验证配置**：先尝试使用 `yolo-v3-tf.json` 配置，如果它能工作，说明问题主要在于后处理参数的不匹配。逐步调整 `yolo-v4-tf.json` 中的参数直到与你的模型输出对齐。\n4. 使用 Netron 等工具查看模型结构，确认输出层的形状和数据类型，以此为依据编写准确的 model-proc 文件。","https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Fissues\u002F220",{"id":160,"question_zh":161,"answer_zh":162,"source_url":163},41082,"如何在同一台机器或 Docker 容器中同时运行多个 DL Streamer 实例处理不同的摄像头流？","要在同一环境中并行运行多个 DL Streamer 管道（例如处理两个不同的摄像头），需注意资源隔离和配置：\n1. **设备指定**：确保每个管道明确指定不同的摄像头设备（如 `\u002Fdev\u002Fvideo0`, `\u002Fdev\u002Fvideo1`）。\n2. **GPU 上下文**：如果在 GPU 上运行，Intel GPU 通常支持多上下文，但需监控显存使用情况。如果显存不足，可能导致第二个实例启动失败。\n3. **Docker 权限**：运行 Docker 时需挂载所有必要的视频设备（`--device \u002Fdev\u002Fvideo0 --device \u002Fdev\u002Fvideo1`）和 DRM 设备（`--device \u002Fdev\u002Fdri`）。\n4. **端口与资源**：如果管道中包含网络接收\u002F发送元素，确保端口不冲突。对于纯本地推理，主要瓶颈通常是 GPU 计算单元或内存带宽。\n5. **性能监控**：使用 `intel_gpu_top` 或 `nvtop` 监控 GPU 利用率，确认硬件是否成为瓶颈。如果单个 GPU 无法支撑多路高负载，考虑使用多卡或降低单路分辨率\u002F帧率。","https:\u002F\u002Fgithub.com\u002Fopen-edge-platform\u002Fdlstreamer\u002Fissues\u002F419",[165,170,175,180,185,190,195,200,205,210,215,220,225,230,235,240,245,250,255,260],{"id":166,"version":167,"summary_zh":168,"released_at":169},324664,"v2026.0.0","# 发布说明：深度学习流媒体管道框架 DL Streamer 2026.0 版\n\n## 版本 2026.0\n\n## 重要亮点：\n\n* 新增元素：gvafpsthrottle、g3dradarprocess、g3dlidarparse\n* 新增模型支持：YOLOv26、YOLO-E、RT-DETR、HuggingFace ViT\n* 简化与 Ultralytics 和 HuggingFace 模型库的集成\n* GstAnalytics 元数据支持：DLStreamer 支持用于目标检测、分类、跟踪的 GstAnalytics 元数据，并新增用于关键点的自定义 GstAnalytics 扩展\n* gvawatermark 全面升级：对象模糊处理、文本背景、标签过滤、更多字体、粗细\u002F颜色选项、FPS 叠加\n* 推理优化：批处理超时机制、适用于所有设备的 OpenCV 张量压缩\n* Windows 平台：通过 D3D11 实现 GPU 推理、gvapython 支持、CI 集成、构建\u002F设置改进\n* 新增 Python 示例：VLM 警报、智能 NVR、ONVIF 设备发现、人脸检测\u002F年龄分类、开放词汇检测、RealSense、DL Streamer + DeepStream\n* 优化器：多流优化、跨流批处理、设备选择，并进行了重构及测试\n* 组件更新：OpenVINO 2026.0.0、NPU 驱动 1.30、RealSense SDK 2.57.5\n* 库整合：将 gvawatermark3d、gvadeskew、gvamotiondetect、gvagenai 合并至 gstvideoanalytics\n* CI：Zizmor 安全扫描、Windows CI、Docker 镜像大小检查\n\n深度学习流媒体管道框架 DL Streamer 是一个基于 GStreamer* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。它确保管道的互操作性，并利用 Intel® Distribution of OpenVINO™ Toolkit 推理引擎后端，在 Intel® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体和推理操作。\n\n该完整解决方案利用以下组件：\n\n- 开源 GStreamer\\* 框架用于管道管理\n- GStreamer* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\n- 视频解码和编码插件，包括 CPU 优化插件或基于 VAAPI 的 GPU 加速插件\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型\n- 以及 Pipeline Framework 仓库中的以下元素：\n\n| 元素 | 描述 |\n|---|---|\n| [gvaattachroi](.\u002Felements\u002Fgvaattachroi.md) | 添加用户自定义感兴趣区域，以便在该区域内进行推理，而非对整个画面进行推理。 |\n| [gvaaudiodetect](.\u002Felements\u002Fgvaaudiodetect.md) | 使用 AclNet 模型执行音频事件检测。 |\n| [gvaaudiotranscribe](.\u002Felements\u002Fgvaaudiotranscribe.md) | 使用 OpenVino GenAI Whisper 模型执行音频转录。 |\n| [gvaclassify](.\u002Felements\u002Fgvaclassify.md) | 执行物体分类。接受 ROI 作为输入，并输出包含 ROI 元数据的分类结果。 |\n| [gvadetect](.\u002Felements\u002Fgvadetect.md) | 使用 YOLOv4-v11、Mobil等目标检测模型，在整个画面或感兴趣区域（ROI）上执行目标检测。 |\n","2026-03-24T09:20:11",{"id":171,"version":172,"summary_zh":173,"released_at":174},324665,"v2025.1.2","## 深度学习流媒体管道框架 2025.1.2 版本\n\n英特尔®深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。它确保管道的互操作性，并利用 Intel® OpenVINO™ 工具套件推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算。\n\n该完整解决方案充分利用以下组件：\n\n- 开源 GStreamer\\* 框架用于管道管理\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\n- 视频解码和编码插件，包括 CPU 优化插件或基于 VAAPI 的 GPU 加速插件\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型\n- 管道框架仓库中的以下元素：\n\n| 元素 | 描述 |\n|---|---|\n| [gvadetect](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvadetect.md) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型，对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标的 ROI。 |\n| [gvaclassify](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaclassify.md) | 执行目标分类。以 ROI 作为输入，输出包含 ROI 元数据的分类结果。 |\n| [gvainference](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvainference.md) | 对全帧或 ROI 运行深度学习推理，支持任何以 RGB 或 BGR 为输入的模型。 |\n| [gvatrack](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvatrack.md) | 使用零样本或无图像跟踪算法执行目标跟踪，为跟踪到的对象分配唯一 ID。 |\n| [gvaaudiodetect](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaaudiodetect.md) | 使用 AclNet 模型执行音频事件检测。 |\n| [gvagenai](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvagenai.md) | 利用 OpenVINO™ GenAI 运行视觉语言模型推理，接受视频和文本提示作为输入，输出文本描述。可用于从视频中生成文本摘要。 |\n| [gvaattachroi](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaattachroi.md) | 添加用户自定义的感兴趣区域，以便在这些区域上而非整个帧上执行推理。 |\n| [gvafpscounter](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvafpscounter.md) | 在单个进程中测量多路流的每秒帧数。 |\n| [gvametaaggregate](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvametaaggregate.md) | 聚合来自多个管道分支的推理结果。 |\n| [gvametaconvert](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvametaconvert.md) | 将元数据结构转换为 JSON 格式。 |\n| [gvametapublish](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvametapublish.md) | 将 JSON 格式的元数据发布到 MQTT 或 Kafka 消息代理，或保存为文件。 |\n| [gvapython](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvapython.md) | 提供回调函数，可在每一帧上执行用户自定义的 Python 函数。","2025-12-19T11:28:11",{"id":176,"version":177,"summary_zh":178,"released_at":179},324666,"v2025.2.0","## 深度学习流媒体管道框架 2025.2.0 版本发布\n\n英特尔®深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析流水线。它确保了管道之间的互操作性，并利用 Intel® OpenVINO™ 工具套件推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算。\n\n该完整解决方案充分利用以下组件：\n\n- 开源 GStreamer\\* 框架用于管道管理\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\n- 视频解码和编码插件，包括 CPU 优化插件和基于 VAAPI 的 GPU 加速插件\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型\n- 管道框架仓库中的以下元素：\n\n| 元素 | 描述 |\n|---|---|\n| [gvaattachroi](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaattachroi.md) | 添加用户自定义感兴趣区域，以便在这些区域内执行推理，而非对整个帧进行处理。 |\n| [gvaaudiodetect](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaaudiodetect.md) | 使用 AclNet 模型执行音频事件检测。 |\n| [gvaaudiotranscribe](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaaudiotranscribe.md) | 使用 OpenVino GenAI Whisper 模型执行音频转录。 |\n| [gvaclassify](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvaclassify.md) | 执行目标分类。接受 ROI 作为输入，并输出包含 ROI 元数据的分类结果。 |\n| [gvadetect](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvadetect.md) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型，在全帧或感兴趣区域（ROI）上执行目标检测。输出检测到的目标的 ROI。 |\n| [gvafpscounter](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvafpscounter.md) | 在单个进程中测量多路流的每秒帧数。 |\n| [gvagenai](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvagenai.md) | 使用 OpenVINO™ GenAI 中的视觉语言模型执行推理，接受视频和文本提示作为输入，并输出文本描述。可用于从视频中生成文本摘要。 |\n| [gvainference](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvainference.md) | 对全帧或 ROI 运行深度学习推理，支持任何以 RGB 或 BGR 输入的模型。 |\n| [gvametaaggregate](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvametaaggregate.md) | 聚合来自多个管道分支的推理结果。 |\n| [gvametaconvert](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvametaconvert.md) | 将元数据结构转换为 JSON 格式。 |\n| [gvametapublish](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvametapublish.md) | 将 JSON 元数据发布到 MQTT 或 Kafka 消息代理，或保存为文件。 |\n| [gvamotiondetect](.\u002Fdocs\u002Fsource\u002Felements\u002Fgvamotiondetect.md) | 对 NV12 格式的视频帧执行轻量级运动检测，并输出运动感兴趣的区域（ROIs）。","2025-12-19T11:29:05",{"id":181,"version":182,"summary_zh":183,"released_at":184},324667,"v2025.0.1.3","# 英特尔®深度学习流媒体管道框架 2025.0.1.3 版本发布\r\n英特尔®深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。该框架确保管道的互操作性，并利用 Intel® OpenVINO™ 工具套件推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算能力。\n\n此版本包含英特尔® DL Streamer 管道框架的相关组件，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\n\n完整解决方案充分利用以下资源：\n\n- 开源的 GStreamer\\* 框架用于管道管理\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\n- 管道框架仓库中的以下组件：\n\n| 组件           | 描述                                                         |\n|----------------|--------------------------------------------------------------|\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标 ROI。 |\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 对全帧或 ROI 运行深度学习推理，支持任何具有 RGB 或 BGR 输入的模型。|\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack) | 使用零样本或无图像跟踪算法执行目标跟踪，并为跟踪到的对象分配唯一 ID。                                                   |\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\n| [gvaattachroi](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaattachroi.html) | 添加用户自定义的感兴趣区域，以便在这些区域内进行推理，而不是在整个帧上进行。|\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | 在单个进程中测量多路流的每秒帧数。 |\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |","2025-04-15T11:58:17",{"id":186,"version":187,"summary_zh":188,"released_at":189},324668,"v2025.0.1.2","# 英特尔® 深度学习流媒体管道框架 2025.0.1.2 版本\r\n英特尔® 深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。它确保管道的互操作性，并利用 Intel® Distribution of OpenVINO™ Toolkit 推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算能力。\r\n\r\n此版本包含英特尔® DL Streamer 管道框架组件，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\r\n\r\n该完整解决方案利用以下技术：\r\n\r\n- 开源 GStreamer\\* 框架进行管道管理\r\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\r\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n- 管道框架仓库中的以下组件：\r\n\r\n| 组件 | 描述 |\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型，在全帧或感兴趣区域（ROI）上执行目标检测。输出检测到的目标的 ROI。  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 在全帧或 ROI 上使用任何支持 RGB 或 BGR 输入的模型运行深度学习推理。|\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| 使用零样本或无图像跟踪算法执行目标跟踪。为跟踪到的目标分配唯一的对象 ID。                                                   |\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\r\n| [gvaattachroi](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaattachroi.html) | 添加用户自定义的感兴趣区域，以便在这些区域而非整个帧上执行推理。|\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | 在单个进程中测量多路流的帧率。 |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |\r\n| ","2025-03-10T15:46:19",{"id":191,"version":192,"summary_zh":193,"released_at":194},324669,"v2025.0.1","# 英特尔®深度学习流媒体管道框架 2025.0.1 版本发布\r\n英特尔®深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。该框架确保管道的互操作性，并利用 Intel® OpenVINO™ 工具套件推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算能力。\r\n\r\n本次发布包含英特尔® DL Streamer 管道框架中的组件，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\r\n\r\n完整解决方案利用以下技术栈：\r\n\r\n- 开源 GStreamer\\* 框架用于管道管理\r\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\r\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n- 管道框架仓库中的以下组件：\r\n\r\n| 组件 | 描述 |\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标 ROI。  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 对全帧或 ROI 运行深度学习推理，支持任何具有 RGB 或 BGR 输入的模型。|\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| 使用零样本或无图像跟踪算法执行目标跟踪，为跟踪到的对象分配唯一 ID。                                                   |\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\r\n| [gvaattachroi](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaattachroi.html) | 添加用户自定义的感兴趣区域，以便在这些区域上而非整个帧上执行推理。|\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | 在单个进程中测量多路流的帧率。 |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |\r\n| [g","2025-02-19T10:54:29",{"id":196,"version":197,"summary_zh":198,"released_at":199},324670,"2025.0.0","# 英特尔® 深度学习流媒体管道框架 2025.0.0 版本\r\n英特尔® 深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。它确保管道的互操作性，并利用 Intel® Distribution of OpenVINO™ Toolkit 推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算功能。\n\n此版本包含英特尔® DL Streamer 管道框架组件，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\n\n该完整解决方案利用以下技术：\n\n- 开源 GStreamer\\* 框架用于管道管理\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\n- 管道框架仓库中的以下组件：\n\n| 组件           | 描述                                                         |\n|----------------|--------------------------------------------------------------|\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标 ROI。 |\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 对全帧或 ROI 运行深度学习推理，支持任何具有 RGB 或 BGR 输入的模型。|\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack) | 使用零样本或无图像跟踪算法执行目标跟踪，为跟踪到的对象分配唯一 ID。                                                   |\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\n| [gvaattachroi](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaattachroi.html) | 添加用户自定义的感兴趣区域，以便在这些区域上执行推理，而不是在整个帧上。|\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | 在单个进程中测量多路流的每秒帧数。 |\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |\n| [g","2025-01-23T09:50:50",{"id":201,"version":202,"summary_zh":203,"released_at":204},324671,"2024.3.0","# 英特尔®深度学习流媒体管道框架 2024.3.0 版本发布\r\n英特尔®深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。该框架确保管道的互操作性，并利用 Intel® OpenVINO™ 工具套件推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算能力。\r\n\r\n本次发布包含英特尔® DL Streamer 管道框架的相关组件，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\r\n\r\n完整解决方案充分利用以下资源：\r\n\r\n- 开源的 GStreamer\\* 框架用于管道管理\r\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\r\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n- 管道框架仓库中的以下组件：\r\n\r\n| 组件           | 描述                                                         |\r\n|----------------|--------------------------------------------------------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型，对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标 ROI。 |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 对全帧或 ROI 运行深度学习推理，支持任何具有 RGB 或 BGR 输入的模型。|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| 使用零样本或无图像跟踪算法进行目标跟踪。为跟踪到的对象分配唯一的对象 ID。                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | 将元数据结构转换为 JSON 格式。|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | 将 JSON 元数据发布到 MQTT 或 Kafka 消息代理，或保存为文件。 |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.githu","2024-12-20T08:41:20",{"id":206,"version":207,"summary_zh":208,"released_at":209},324672,"v2024.2.2","# 英特尔® 深度学习流媒体管道框架 2024.2.2 版本发布\r\n英特尔® 深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。该框架确保管道的互操作性，并利用 Intel® Distribution of OpenVINO™ Toolkit 推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算能力。\n\n本次发布包含英特尔® DL Streamer 管道框架中的多个元素，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\n\n完整解决方案充分利用以下技术：\n\n- 开源的 GStreamer\\* 框架用于管道管理\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\n- 管道框架仓库中的以下元素：\n\n| 元素 | 描述 |\n|--------|------------|\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect) | 使用 YOLOv4-v11、MobileNet SSD、Faster-RCNN 等目标检测模型对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标 ROI。 |\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 对全帧或 ROI 运行深度学习推理，支持任何具有 RGB 或 BGR 输入的模型。|\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| 使用零样本或无图像跟踪算法进行目标跟踪。为跟踪到的对象分配唯一的对象 ID。                                                   |\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | 将元数据结构转换为 JSON 格式。|\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | 将 JSON 元数据发布到 MQTT 或 Kafka 消息代理，或保存为文件。 |\n| [gvapython](https:\u002F\u002Fdlstreamer.githu","2024-11-29T13:19:58",{"id":211,"version":212,"summary_zh":213,"released_at":214},324673,"v2024.2.1","# 英特尔® 深度学习流媒体管道框架 2024.2.1 版本\r\n英特尔® 深度学习流媒体（Intel® DL Streamer）管道框架是一个基于 GStreamer\\* 多媒体框架的流媒体分析框架，用于构建复杂的媒体分析管道。该框架确保管道的互操作性，并利用 Intel® Distribution of OpenVINO™ Toolkit 推理引擎后端，在英特尔® 架构、CPU、独立 GPU、集成 GPU 和 NPU 上提供优化的媒体处理和推理运算能力。\r\n\r\n本次发布包含英特尔® DL Streamer 管道框架中的组件，以实现视频和音频分析功能（例如目标检测、分类、音频事件检测），以及其他用于在 GStreamer\\* 框架中构建端到端优化管道的组件。\r\n\r\n完整解决方案充分利用以下技术：\r\n\r\n- 开源的 GStreamer\\* 框架用于管道管理\r\n- GStreamer\\* 插件用于输入和输出，例如媒体文件以及来自摄像头或网络的实时流\r\n- 视频解码和编码插件，包括 CPU 优化插件和 GPU 加速插件[基于 VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n- 从 TensorFlow\\*、Caffe\\* 等训练框架转换而来的深度学习模型，这些模型来自 [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n- 管道框架仓库中的以下组件：\r\n\r\n| 组件 | 描述 |\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect) | 使用 YOLOv4、MobileNet SSD、Faster-RCNN 等目标检测模型对全帧或感兴趣区域（ROI）进行目标检测。输出检测到的目标 ROI。 |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | 执行目标分类。接受 ROI 作为输入，并输出带有 ROI 元数据的分类结果。                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | 对全帧或 ROI 运行深度学习推理，支持任何具有 RGB 或 BGR 输入的模型。|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | 使用 AclNet 模型执行音频事件检测。 |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| 使用零样本或无图像跟踪算法进行目标跟踪。为跟踪到的目标分配唯一的对象 ID。                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | 聚合来自多个管道分支的推理结果 |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | 将元数据结构转换为 JSON 格式。|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | 将 JSON 元数据发布到 MQTT 或 Kafka 消息代理，或保存为文件。 |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io","2024-10-31T09:48:32",{"id":216,"version":217,"summary_zh":218,"released_at":219},324674,"v2024.2.0","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.2.0\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discrete GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, instance segmentation, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n|[gvaattachroi](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaattachroi)|Provides an ability to define one or more regions of interest to perform inference on, instead of the full frame. |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n|New models support: Yolov10 for GPU, DeepLabv3 | Support for most recent Yolov10 model for GPU and DeepLabv3 (semantic segmentation) |\r\n|UTC format for timestamp| Timestamp can be shown in UTC format based on system time with option to synchronize it from NTP server |\r\n|OpenVINO 2024.4 support | Update to latest version of OpenVINO |\r\n|GStreamer 1.24.7 support | Update to latest version of GStreamer |\r\n|Intel® NPU 1.6.0 driver support | Support for newer version of Intel® NPU Linux driver |\r\n|Simplified installation process for option#1 (i.e. Ubuntu packages) via script|Development of the script that enhances user experience during installation of Intel® DL Streamer with usage of option#1. |\r\n|Documentation improvements|Descriptions enhancements in various points.|\r\n[Preview fea","2024-09-30T15:28:24",{"id":221,"version":222,"summary_zh":223,"released_at":224},324675,"v2024.1.2","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.1.2\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discrete GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, instance segmentation, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n|[gvaattachroi](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaattachroi)|Provides an ability to define one or more regions of interest to perform inference on, instead of the full frame. |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n|New models support: Yolov10 for CPU only,Yolov8 instance segmentation | Support for most recent Yolov10 model for CPU and extension for Yolov8 |\r\n|New elements: gvaattachroi including documentation update + samples) | Added element documentation and sample development which introduces ability to define the area of interest on which the inference should be performed |\r\n|OpenVINO 2024.3 support | Update to latest version of OpenVINO |\r\n|GStreamer 1.24.6 support | Update to latest version of GStreamer |\r\n|Ubuntu 24.04 support | Support for newer version of Ubuntu |\r\n|Documentation updates for DeepStream to DL Streamer migration process | Updates to the migration process from Deep Stream |\r\n|Documentation improvements | Descriptions enhancements in various points |\r\n[Preview feature","2024-08-30T09:38:58",{"id":226,"version":227,"summary_zh":228,"released_at":229},324676,"v2024.1.1","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.1.1\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discrete GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Missing git package | Git package added to DLStreamer docker runtime image |\r\n| VTune when running DLStreamer | Publish instructions to install and run VTune to analyze media + gpu when running DLStreamer  |\r\n| Update NPU drivers to version 1.5.0 | Update NPU driver version inside docker images|\r\n| Instance_segmentation sample | Add new Mask-RCNN segmentation sample |\r\n| Documentation updates | Enhance Performance Guide and Model Preparation section |\r\n| Fix samples errors | Fixed errors on action_recognition, geti, yolo and ffmpeg (customer issue) samples |\r\n| Fix memory grow with `meta_overlay` | Fix for Meta Overlay memory leak with DLS Arch 2.0 |\r\n| Fix pipeline which failed to start with mobilenet-v2-1.0-224 model  | \r\n| Fix batch-size error -> with yolov8 model and other yolo models | \r\n\r\n\r\n## Known Issues\r\n\r\n| **Issue**   | **Issue Description**  |\r\n|----------------|------------------------|\r\n| VAAPI memory with `decodebin` | If you are using `decodebin` in conjunction with","2024-07-29T11:00:49",{"id":231,"version":232,"summary_zh":233,"released_at":234},324677,"v2024.1.0","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.1.0\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discrete GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Switch to ‘gst-va’ as default processing path instead of ‘gst-vaapi’ | Switch to ‘gst-va’ as default processing path instead of ‘gst-vaapi’ |\r\n| Add support for ‘gst-qsv’ plugins | Add support for ‘qsv’ plugins |\r\n| New public ONNX models: Centerface and HSEmotion | New public ONNX models: Centerface and HSEmotion |\r\n| Update Gstreamer version to the latest one (current 1.24) | Update Gstreamer version to the latest one (1.24.4) |\r\n| Update OpenVINO version to latest one (2024.2.0) | Update OpenVINO version to latest one (2024.2.0) |\r\n| Release docker images on DockerHUB: runtime and dev | Release docker images on DockerHUB: runtime and dev |\r\n| Bugs fixing | Bug fixed: GPU not detected in Docker container Dlstreamer - MTL platform; Updated docker images with proper GPU and NPU packages; yolo5 model failed with batch-size >1; Remove excessive ‘mbind failed:...’ warning logs |\r\n| Documentation updates | Added sample applications for Mask-RCNN instance segmentation. Added list of sup","2024-06-27T12:11:28",{"id":236,"version":237,"summary_zh":238,"released_at":239},324678,"v2024.0.2","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.0.2\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discreate GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Support for ‘gst-va’ in addition to ‘gst-vaapi’ | Support for ‘gst-va’ in addition to ‘gst-vaapi’ |\r\n| Add support for EfficentNetv2 (classification), MaskRCNN (instance segmentation) and Yolo8-OBB (oriented bounding box) | New classification model supported EfficentNetv2, new instance segmentation model supported MaskRCNN and oriented bounding box model as well added Yolo8-OBB |\r\n| Support additional GETI models: segmentation, obb | GETI public models support added |\r\n| Generalized method to deploy new models without need for model-proc file | Support model information embedded into AI model descriptors according to OpenVINO Model API |\r\n| Release docker images on DockerHUB: runtime | Added docker images on DockerHUB: runtime |\r\n| Added support for OpenVINO 2024.1.0 | Added support for OpenVINO 2024.1.0 |\r\n\r\n## Acknowledgements\r\nThanks for contributions from the DL Streamer developer community:\r\n@aminatef\r\n@russkel\r\n\r\n## Fixed issues\r\n| **Issue \\#**   | **Issue Description**  | *","2024-05-29T16:07:27",{"id":241,"version":242,"summary_zh":243,"released_at":244},324679,"v2024.0.1","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.0.1\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discreate GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Add support for latest Ultralytics YOLO models | Add support for latest Ultralytics YOLO models: -v7, -v8, -v9 |\r\n| Add support for YOLOX models | Add support for YOLOX models |\r\n| Support deployment of GETI-trained models | Support models trained by GETI v1.8: bounding-box detection and classification (single and multi-label) |\r\n| Automatic pre-\u002Fpost-processing based on model descriptor | Automatic pre-\u002Fpost-processing based on model descriptor (model-proc file not required): yolov8, yolov9 and GETI |\r\n| Docker image size reduction | Reduced docker image size generated from the published docker file |\r\n\r\n## Changed in this Release\r\n\r\n### Docker image replaced with Docker file\r\n* Ubuntu 22.04 reduced docker file is released.\r\n\r\n## Known Issues\r\n\r\n| **Issue**   | **Issue Description**  |\r\n|----------------|------------------------|\r\n| VAAPI memory with `decodebin` | If you are using `decodebin` in conjunction with `vaapi-surface-sharing` preprocessing backend you should set caps fi","2024-04-25T08:21:29",{"id":246,"version":247,"summary_zh":248,"released_at":249},324680,"v2024.0","# Intel® Deep Learning Streamer Pipeline Framework Release 2024.0\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, discreate GPU, integrated GPU and NPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Intel® Core™ Ultra processors NPU support | Inference on NPU devices has been added, validated with Intel(R) Core(TM) Ultra 7 155H |\r\n| Compatibility with OpenVINO™ Toolkit 2024.0 | Pipeline Framework has been updated to use the 2024.0.0 version of the OpenVINO™ Toolkit |\r\n| Compatibility with GStreamer 1.22.9 | Pipeline Framework has been updated to use GStreamer framework version 1.22.9 |\r\n| Updated to FFmpeg 6.1.1 | Updated FFmpeg from 5.1.3 to 6.1.1 |\r\n| Performance optimizations | 8% geomean gain across tested scenarios, up to 50% performance gain in multi-stream scenarios |\r\n\r\n## Changed in this Release\r\n\r\n### Docker image replaced with Docker file\r\n* Ubuntu 22.04 docker file is released instead of docker image.\r\n\r\n## Known Issues\r\n\r\n| **Issue**   | **Issue Description**  |\r\n|----------------|------------------------|\r\n| Intermittent accuracy fails with YOLOv5m and YOLOv5s | Object detection pipelines using YOLOv5m and YOLOv5s show intermittent inconstancy between runs |\r\n| VA","2024-03-27T13:52:06",{"id":251,"version":252,"summary_zh":253,"released_at":254},324681,"2023.0-release","# Intel® Deep Learning Streamer Pipeline Framework Release 2023.0\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, and iGPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaaggregate.html) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Compatibility with OpenVINO™ Toolkit 2023.0 | Pipeline Framework has been updated to use the 2023.0.0 version of the OpenVINO™ Toolkit |\r\n| Intel® Data Center GPU Flex Series PV support | Validated on Intel® Data Center GPU Flex Series [140 and 170](https:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fproducts\u002Fdocs\u002Fdiscrete-gpus\u002Fdata-center-gpu\u002Fflex-series\u002Fproduct-brief.html) with pipelines\u002Fmodels\u002Fvideos from the Intel® DL Streamer [Pipeline Zoo](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo), [Pipeline Zoo Models](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo-models) and [Pipeline Zoo Media](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo-media) repositories.  Tested with the Latest GPU Linux release (https:\u002F\u002Fdgpu-docs.intel.com\u002Freleases\u002Fproduction_682.14_20230804.html) |\r\n| Updated to FFmpeg 5.1.3 | Updated FFmpeg from 5.1 to 5.1.3 |\r\n| New media analytics model support | Added support for DeepSort and object tracking |\r\n\r\n## Changed in this Release\r\n\r\n### Deprecation Notices\r\n* Ubuntu 20.04 is no longer actively supported.  \r\n* See","2023-10-02T18:24:33",{"id":256,"version":257,"summary_zh":258,"released_at":259},324682,"2022.3-release","# Intel® Deep Learning Streamer Pipeline Framework Release 2022.3\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, and iGPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgitlab.freedesktop.org\u002Fgstreamer\u002Fgstreamer\u002F-\u002Ftree\u002Fmain\u002Fsubprojects\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLOv4, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](dlstreamer.github.io\u002Felements\u002Fgvametaaggregate) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Intel® Data Center GPU Flex Series PV support | Validated on Intel® Data Center GPU Flex Series [140 and 170](https:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fproducts\u002Fdocs\u002Fdiscrete-gpus\u002Fdata-center-gpu\u002Fflex-series\u002Fproduct-brief.html) with pipelines\u002Fmodels\u002Fvideos from the Intel® DL Streamer [Pipeline Zoo](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo), [Pipeline Zoo Models](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo-models) and [Pipeline Zoo Media](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo-media) repositories |\r\n| Full Ubuntu 22.04 Support | Intel® DL Streamer has moved primary support to the current Ubuntu 22.04 LTS release.  Ubuntu 20.04 is still a supported OS but Docker Images and APT packages are based on 22.04  |\r\n| Compatibility with OpenVINO™ Toolkit 2022.3 | Pipeline Framework has been updated to use the 2022.3.0 version of the OpenVINO™ Toolkit |\r\n| Updated to FFmpeg 5.1 | Updated FFmpeg from 4.4 to 5.1 |\r\n\r\n## Changed in this Release\r\n\r\n### Deprecation Notices\r\n\r\n* Ubuntu 20.04 is still supported but primary support has mov","2023-03-03T21:51:39",{"id":261,"version":262,"summary_zh":263,"released_at":264},324683,"2022.2-release","# Intel® Deep Learning Streamer Pipeline Framework 2022.2\r\n\r\nIntel® Deep Learning Streamer (Intel® DL Streamer) Pipeline Framework is a streaming media analytics framework, based on GStreamer\\* multimedia framework, for creating complex media analytics pipelines. It ensures pipeline interoperability and provides optimized media, and inference operations using Intel® Distribution of OpenVINO™ Toolkit Inference Engine backend, across Intel® architecture, CPU, and iGPU.\r\n\r\nThis release includes Intel® DL Streamer Pipeline Framework elements to enable video and audio analytics capabilities, (e.g., object detection, classification, audio event detection), and other elements to build end-to-end optimized pipeline in GStreamer\\* framework.\r\n\r\nThe complete solution leverages:\r\n\r\n-   Open source GStreamer\\* framework for pipeline management\r\n-   GStreamer\\* plugins for input and output such as media files and real-time streaming from camera or network\r\n-   Video decode and encode plugins, either CPU optimized plugins or GPU-accelerated plugins [based on VAAPI](https:\u002F\u002Fgithub.com\u002FGStreamer\u002Fgstreamer-vaapi)\r\n-   Deep Learning models converted from training frameworks TensorFlow\\*, Caffe\\* etc. from [Open Model Zoo (OMZ)](https:\u002F\u002Fgithub.com\u002Fopenvinotoolkit\u002Fopen_model_zoo)\r\n-   The following elements in the Pipeline Framework repository:\r\n\r\n| Element| Description|\r\n|--------|------------|\r\n| [gvadetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvadetect)| Performs object detection on a full-frame or region of interest (ROI) using object detection models such as YOLO v3-v5, MobileNet SSD, Faster-RCNN etc. Outputs the ROI for detected objects.  |\r\n| [gvaclassify](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaclassify) | Performs object classification. Accepts the ROI as an input and outputs classification results with the ROI metadata.                                                                      |\r\n| [gvainference](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvainference) | Runs deep learning inference on a full-frame or ROI using any model with an RGB or BGR input.|\r\n| [gvaaudiodetect](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvaaudiodetect) | Performs audio event detection using AclNet model. |\r\n| [gvatrack](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvatrack)| Performs object tracking using zero-term, or imageless tracking algorithms. Assigns unique object IDs to the tracked objects.                                                   |\r\n| [gvametaaggregate](dlstreamer.github.io\u002Felements\u002Fgvametaaggregate) | Aggregates inference results from multiple pipeline branches |\r\n| [gvametaconvert](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametaconvert) | Converts the metadata structure to the JSON format.|\r\n| [gvametapublish](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvametapublish) | Publishes the JSON metadata to MQTT or Kafka message brokers or files. |\r\n| [gvapython](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvapython) | Provides a callback to execute user-defined Python functions on every frame. Can be used for metadata conversion, inference post-processing, and other tasks.|\r\n| [gvawatermark](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvawatermark) | Overlays the metadata on the video frame to visualize the inference results. |\r\n| [gvafpscounter](https:\u002F\u002Fdlstreamer.github.io\u002Felements\u002Fgvafpscounter) | Measures frames per second across multiple streams in a single process |\r\n\r\nFor the details of supported platforms, please refer to [System Requirements](#system-requirements) section.\r\n\r\nFor installing Pipeline Framework with the prebuilt binaries or Docker\\* or to build the binaries from the open source, please refer to [Intel® DL Streamer Pipeline Framework installation guide](https:\u002F\u002Fdlstreamer.github.io\u002Fget_started\u002Finstall\u002Finstall_guide_index.html)\r\n\r\n## New in this Release\r\n\r\n| **Title**      | **High-level description**      |\r\n|----------------|---------------------------------|\r\n| Intel® Data Center GPU Flex Series Beta support | Validated on Intel® Data Center GPU Flex Series [140 and 170](https:\u002F\u002Fwww.intel.com\u002Fcontent\u002Fwww\u002Fus\u002Fen\u002Fproducts\u002Fdocs\u002Fdiscrete-gpus\u002Fdata-center-gpu\u002Fflex-series\u002Fproduct-brief.html) with pipelines\u002Fmodels\u002Fvideos from the Intel® DL Streamer [Pipeline Zoo](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo), [Pipeline Zoo Models](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo-models) and [Pipeline Zoo Media](https:\u002F\u002Fgithub.com\u002Fdlstreamer\u002Fpipeline-zoo-media) repositories |\r\n| Updated to GStreamer 1.20.3 | Upgrades from GStreamer 1.18.4 to latest stable GStreamer 1.20.3. |\r\n| YOLOv5 Support | Added YOLOv5 postprocessing support |\r\n| Architecture 2.0 [Preview] | Includes memory interop header-only library for zero-copy buffer sharing on CPU and GPU, C++ elements, integration into GStreamer as [three sub-components](https:\u002F\u002Fdlstreamer.github.io\u002Farchitecture_2.0\u002Farchitecture_2.0.html) |\r\n| New non-GStreamer samples 2.0 | Samples [FFmpeg+OpenVINO and FFmpeg+DPCPP](https:\u002F\u002Fdlstreamer.github.io\u002Farchitecture_2.0\u002Fsamples_2.0.html) |\r\n| New GStreamer ","2022-10-07T00:55:10"]