OpenPose Rebuilt-Python

Rebuilting the CMU-OpenPose pose estimatior using Python with OpenCV and Tensorflow.
(The code comments are partly descibed in chinese)

Pretrained-model Downloading

In this work, I used both caffemodel and tensorflow-graph-model, you can download them here, Then place the pretrained models to corresponding directory respectively.

Examples:

• place caffe_modelsposebody_25pose_iter_584000.caffemodel into pose-estimator-using-caffemodelmodelbody_25

• place caffe_modelshandpose_iter_102000.caffemodel into hand-estimator-using-caffemodelmodel

• place openpose graph model cocograph_opt.pb into pose-estimator-tensorflowgraph_model_coco

Requirements :

• OpenCV > 3.4.1

• TensorFlow > 1.2.0

• imutils

Usage:

See the sub-README.md in sub-folder.

BODY_25 vs. COCO vs. MPI

• BODY_25 model is faster, more accurate, and it includes foot keypoints.

• COCO requires less memory on GPU (being able to fit into 2GB GPUs with the default settings) and it runs faster on CPU-only mode.

• MPI model is only meant for people requiring the MPI-keypoint structure. It is also slower than BODY_25 and far less accurate.

Output Format

Body_25 in left, COCO in middle, MPI in right.

See more Output Format details here, and Hand Output Format included as well.