MMPose Support¶
MMPose is an open-source toolbox for pose estimation based on PyTorch. It is a part of the OpenMMLab project.
MMPose installation tutorial¶
Please refer to official installation guide to install the codebase.
MMPose models support¶
| Model | Task | ONNX Runtime | TensorRT | ncnn | PPLNN | OpenVINO | Model config |
|---|---|---|---|---|---|---|---|
| HRNet | PoseDetection | Y | Y | Y | N | Y | config |
| MSPN | PoseDetection | Y | Y | Y | N | Y | config |
| LiteHRNet | PoseDetection | Y | Y | Y | N | Y | config |
Example¶
python tools/deploy.py \
configs/mmpose/posedetection_tensorrt_static-256x192.py \
$MMPOSE_DIR/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/hrnet_w48_coco_256x192.py \
$MMPOSE_DIR/checkpoints/hrnet_w48_coco_256x192-b9e0b3ab_20200708.pth \
$MMDEPLOY_DIR/demo/resources/human-pose.jpg \
--work-dir work-dirs/mmpose/topdown/hrnet/trt \
--device cuda
Note
Usually, mmpose models need some extra information for the input image, but we can’t get it directly. So, when exporting the model, you can use
$MMDEPLOY_DIR/demo/resources/human-pose.jpgas input.