本文将疏导疾速应用 MMSkeleton ，介绍用摄像头测试实时姿势预计。

  • MMSkeleton: https://github.com/open-mmlab...

装置

首先装置 MMDetection ，可见 MMDetection 应用。

而后装置 MMSkeleton ，

# 启用 Python 虚拟环境conda activate open-mmlab# 下载 MMSkeletongit clone https://github.com/open-mmlab/mmskeleton.gitcd mmskeleton# 装置 MMSkeletonpython setup.py develop# 装置 nms op for person estimationcd mmskeleton/ops/nms/python setup_linux.py developcd ../../../

现有模型，视频测试

配置

configs/pose_estimation/pose_demo.yaml:

processor_cfg:  video_file: resource/data_example/ta_chi.mp4  detection_cfg:    model_cfg: ../mmdetection/configs/cascade_rcnn/cascade_rcnn_r50_fpn_1x_coco.py    checkpoint_file: ../mmdetection/checkpoints/cascade_rcnn_r50_fpn_1x_coco_20200316-3dc56deb.pth    bbox_thre: 0.8

选用的检测模型，如下：

  • Cascade R-CNN, R-50-FPN, 1x

    • config
    • model

运行

# verify that mmskeleton and mmdetection installed correctly# python mmskl.py pose_demo [--gpus $GPUS]python mmskl.py pose_demo --gpus 1

后果将会存到 work_dir/pose_demo/ta_chi.mp4。

现有模型，WebCam 测试

配置

configs/apis/pose_estimator.cascade_rcnn+hrnet.yaml:

detection_cfg:  model_cfg: mmdetection/configs/cascade_rcnn/cascade_rcnn_r50_fpn_1x_coco.py  checkpoint_file: mmdetection/checkpoints/cascade_rcnn_r50_fpn_1x_coco_20200316-3dc56deb.pth  bbox_thre: 0.8estimation_cfg:  model_cfg: mmskeleton/configs/pose_estimation/hrnet/pose_hrnet_w32_256x192_test.yaml  checkpoint_file: mmskeleton://pose_estimation/pose_hrnet_w32_256x192  data_cfg:    image_size:      - 192      - 256    pixel_std: 200    image_mean:      - 0.485      - 0.456      - 0.406    image_std:      - 0.229      - 0.224      - 0.225    post_process: true

确认 detection_cfg estimation_cfg 的门路正确。

写码

编写 webcam.py，次要代码如下：

def main():  args = parse_args()  win_name = args.win_name  cv.namedWindow(win_name, cv.WINDOW_NORMAL)  with Camera(args.cam_idx, args.cam_width, args.cam_height, args.cam_fps) as cam:    cfg = mmcv.Config.fromfile(args.cfg_file)    detection_cfg = cfg["detection_cfg"]    print("Loading model ...")    model = init_pose_estimator(**cfg, device=0)    print("Loading model done")    for frame in cam.reads():      res = inference_pose_estimator(model, frame)      res_image = pose_demo.render(          frame, res["joint_preds"], res["person_bbox"],          detection_cfg.bbox_thre)      cv.imshow(win_name, res_image)      key = cv.waitKey(1) & 0xFF      if key == 27 or key == ord("q"):        break  cv.destroyAllWindows()

运行

$ python webcam.py \--cam_idx 2 --cam_width 640 --cam_height 480 --cam_fps 10 \--cfg_file configs/apis/pose_estimator.cascade_rcnn+hrnet.yamlArgs  win_name: webcam  cam_idx: 2  cam_width: 640  cam_height: 480  cam_fps: 10  cfg_file: configs/apis/pose_estimator.cascade_rcnn+hrnet.yamlCAM: 640.0x480.0 10.0Loading model ...Loading model done

成果，

摄像头参数，可见 WebCam 摄像头应用。

更多

  • Awesome Human Pose Estimation
  • Awesome Skeleton based Action Recognition
GoCoding 集体实际的教训分享，可关注公众号！