RTMPose, a state-of-the-art framework developed by Shanghai AI Laboratory, excels in real-time multi-person pose estimation by integrating an innovative model architecture with the efficiency of the MMPose foundation. The framework's architecture is meticulously designed to enhance performance and reduce latency, making it suitable for a variety of applications where real-time analysis is crucial.
| GPU | IXUCA SDK | Release | Branch |
|---|---|---|---|
| MR-V100 | 4.4.0 | 26.03 | release/26.03 |
| MR-V100 | 4.3.0 | 25.12 | release/25.12 |
Note: 请切换到与您的 SDK 版本对应的 Release 分支进行测试。请勿直接在 master 分支上运行测试，因为 master 分支可能包含与您的本地 SDK 版本不兼容的最新更改。
切换分支命令示例：
git checkout release/26.03
Pretrained model: https://download.openmmlab.com/mmpose/v1/projects/rtmposev1/rtmpose-m_simcc-aic-coco_pt-aic-coco_420e-256x192-63eb25f7_20230126.pth
Dataset:
- https://github.com/ultralytics/assets/releases/download/v0.0.0/coco2017labels.zip to download the labels dataset.
- http://images.cocodataset.org/zips/val2017.zip to download the validation dataset.
- http://images.cocodataset.org/zips/train2017.zip to download the train dataset.
- http://images.cocodataset.org/annotations/annotations_trainval2017.zip to download the annotations dataset.
unzip -q -d ./ coco2017labels.zip
unzip -q -d ./coco/images/ train2017.zip
unzip -q -d ./coco/images/ val2017.zip
unzip -q -d ./coco annotations_trainval2017.zip
coco
├── annotations
│ └── instances_train2017.json
│ └── instances_val2017.json
│ └── captions_train2017.json
│ └── captions_val2017.json
│ └── person_keypoints_train2017.json
│ └── person_keypoints_val2017.json
├── images
│ ├── train2017
│ └── val2017
├── labels
│ ├── train2017
│ └── val2017
├── LICENSE
├── README.txt
├── test-dev2017.txt
├── train2017.cache
├── train2017.txt
├── val2017.cache
└── val2017.txtwget https://download.openmmlab.com/mmpose/v1/projects/rtmposev1/rtmpose-m_simcc-aic-coco_pt-aic-coco_420e-256x192-63eb25f7_20230126.pth# Install libGL
## CentOS
yum install -y mesa-libGL
## Ubuntu
apt install -y libgl1-mesa-glx
# before install mmpose==1.3.1 need to install chchumpy==0.70 which is too older that is not compatible with newer Python versions or pip
# so need to downgrade pip to version 20.2.4
pip install pip==20.2.4
pip install mmpose==1.3.1
pip install --upgrade pip
pip install -r requirements.txt# set weights_only=False to be comaptible with pytorch 2.7
sed -i '347 s/map_location)/map_location, weights_only=False)/' /usr/local/lib/python3.10/site-packages/mmengine/runner/checkpoint.py
# export onnx model
python3 export.py --weight rtmpose-m_simcc-aic-coco_pt-aic-coco_420e-256x192-63eb25f7_20230126.pth --cfg rtmpose-m_8xb256-420e_coco-256x192.py --output rtmpose.onnx
# use onnxsim optimize onnx model
onnxsim rtmpose.onnx rtmpose_opt.onnxexport DATASETS_DIR=/Path/to/coco/# Accuracy
bash scripts/infer_rtmpose_fp16_accuracy.sh
# Performance
bash scripts/infer_rtmpose_fp16_performance.sh| Model | BatchSize | Precision | FPS | IOU@0.5 | IOU@0.5:0.95 |
|---|---|---|---|---|---|
| RTMPose | 32 | FP16 | 2313.33 | 0.936 | 0.773 |