diff --git a/deploy/pptracking/README.md b/deploy/pptracking/README.md new file mode 100644 index 0000000000000000000000000000000000000000..3b9944a9b759ecbb5fe15829979cb5eeb438e5c5 --- /dev/null +++ b/deploy/pptracking/README.md @@ -0,0 +1,46 @@ +# 实时跟踪系统PP-Tracking + +PP-Tracking是基于飞桨深度学习框架的业界首个开源实时跟踪系统。针对实际业务的难点痛点，PP-Tracking内置行人车辆跟踪、跨镜头跟踪、多类别跟踪、小目标跟踪及流量计数等能力与产业应用，同时提供可视化开发界面。模型集成多目标跟踪，目标检测，ReID轻量级算法，进一步提升PP-Tracking在服务器端部署性能。同时支持python，C++部署，适配Linux，Nvidia Jetson多平台环境。 + +
+ +
+ +
+ +
+ +### 一、快速开始 + +PP-Tracking提供了简洁的可视化界面，无需开发即可实现多种跟踪功能，可以参考[PP-Tracking可视化界面使用文档]()快速上手体验 + +### 二、算法介绍 + +PP-Tracking集成了多目标跟踪，目标检测，ReID轻量级算法，提升跟踪系统实时性能。多目标跟踪算法基于FairMOT进行优化，实现了服务器端轻量级模型，同时基于不同应用场景提供了针对性的预训练模型。 + +模型训练评估方法请参考[多目标跟踪快速开始](../../configs/mot/README_cn.md#快速开始) + +PP-Tracking中提供的多场景预训练模型及导出模型列表如下： + +| 场景 | 数据集 | 精度（MOTA） | NX模型预测速度（FPS） | 配置文件 | 模型权重 | 预测部署模型 | +| :---------:|:--------------- | :-------: | :------: | :------: |:---: | :---: | +| 行人跟踪 | MOT17 | 65.3 | - | [配置文件](../../configs/mot/fairmot/fairmot_hrnetv2_w18_dlafpn_30e_576x320.yml) | [下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_576x320.pdparams) | [下载链接](https://bj.bcebos.com/v1/paddledet/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_576x320.tar) | +| 行人小目标跟踪 | VisDrone-pedestrian | 40.5 | -| [配置文件](../../configs/mot/pedestrian/fairmot_hrnetv2_w18_dlafpn_30e_864x480_visdrone_pedestrian.yml) | [下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_864x480_visdrone_pedestrian.pdparams) | [下载链接](https://bj.bcebos.com/v1/paddledet/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_1088x608_visdrone_pedestrian.tar) | +| 车辆跟踪 | BDD100k-vehicle | 32.6 | - | [配置文件](../../configs/mot/vehicle/fairmot_hrnetv2_w18_dlafpn_30e_576x320_bdd100kmot_vehicle.yml) | [下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_576x320_bdd100kmot_vehicle.pdparams) | [下载链接](https://bj.bcebos.com/v1/paddledet/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_576x320_bdd100kmot_vehicle.tar) | +| 车辆小目标跟踪 | VisDrone-vehicle | 39.8 | - | [配置文件](../../configs/mot/vehicle/fairmot_hrnetv2_w18_dlafpn_30e_576x320_visdrone_vehicle.yml) | [下载链接](https://paddledet.bj.bcebos.com/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_576x320_visdrone_vehicle.pdparams) | [下载链接](https://bj.bcebos.com/v1/paddledet/models/mot/fairmot_hrnetv2_w18_dlafpn_30e_576x320_visdrone_vehicle.tar) +| 多类别跟踪 | BDD100k | - | - | [配置文件]() | [下载链接]() | [下载链接](https://bj.bcebos.com/v1/paddledet/models/mot/mcfairmot_hrnetv2_w18_dlafpn_30e_576x320_bdd100k_mcmot.tar) | +| 多类别小目标跟踪 | VisDrone | 20.4 | - | [配置文件](../../configs/mot/mcfairmot/mcfairmot_hrnetv2_w18_dlafpn_30e_1088x608_visdrone.yml) | [下载链接](https://paddledet.bj.bcebos.com/models/mot/mcfairmot_hrnetv2_w18_dlafpn_30e_576x320_visdrone.pdparams) | [下载链接](https://bj.bcebos.com/v1/paddledet/models/mot/mcfairmot_hrnetv2_w18_dlafpn_30e_1088x608_visdrone.tar) | + +更多跟踪模型请参考[多目标跟踪模型库](../../configs/mot/README_cn.md#模型库) + +检测模型使用轻量级特色模型PP-PicoDet，具体请参考[PP-PicoDet文档](../../configs/picodet) + +ReID模型使用超轻量骨干网络模型PP-LCNet, 具体请参考[PP-LCNet模型介绍](https://github.com/PaddlePaddle/PaddleClas/blob/release/2.3/docs/zh_CN/models/PP-LCNet.md) + +### 三、Python端预测部署 + +PP-Tracking 使用python预测部署教程请参考[PP-Tracking python部署文档](python/README.md) + +### 四、C++端预测部署 + +PP-Tracking 使用c++预测部署教程请参考[PP-Tracking c++部署文档](cpp/README.md) diff --git a/deploy/pptracking/python/README.md b/deploy/pptracking/python/README.md index c086d8df1cd46dcb04c69c81e050ca945c744036..707b66c4a65739155e061e651ad7de1a212cefed 100644 --- a/deploy/pptracking/python/README.md +++ b/deploy/pptracking/python/README.md @@ -86,6 +86,8 @@ python deploy/pptracking/python/mot_sde_infer.py --model_dir=output_inference/pp | --enable_mkldnn | Option | CPU预测中是否开启MKLDNN加速，默认为False | | --cpu_threads | Option| 设置cpu线程数，默认为1 | | --trt_calib_mode | Option| TensorRT是否使用校准功能，默认为False。使用TensorRT的int8功能时，需设置为True，使用PaddleSlim量化后的模型时需要设置为False | +| --do_entrance_counting | Option | 是否统计出入口流量，默认为False | +| --draw_center_traj | Option | 是否绘制跟踪轨迹，默认为False | 说明： diff --git a/docs/images/pptracking-demo.gif b/docs/images/pptracking-demo.gif new file mode 100644 index 0000000000000000000000000000000000000000..8b8cbd5983163e979dbd63e340a70aa03853da70 Binary files /dev/null and b/docs/images/pptracking-demo.gif differ diff --git a/docs/images/pptracking.png b/docs/images/pptracking.png new file mode 100644 index 0000000000000000000000000000000000000000..93ae946004bb1f729ee814348945eeb58ce70ebe Binary files /dev/null and b/docs/images/pptracking.png differ