docs/inference.md

In-the-wild Inference

2D Pose

Please use AlphaPose to extract the 2D keypoints for your video first. We use the Fast Pose model trained on Halpe dataset (Link).

Note: Currently we only support single person. If your video contains multiple person, you may need to use the Pose Tracking Module for AlphaPose and set --focus to specify the target person id.

3D Pose

1. Please download the checkpoint here and put it to checkpoint/pose3d/FT_MB_lite_MB_ft_h36m_global_lite/.
2. Run the following command to infer from the extracted 2D poses:

python infer_wild.py \
--vid_path <your_video.mp4> \
--json_path <alphapose-results.json> \
--out_path <output_path>

Mesh

1. Please download the checkpoint here and put it to checkpoint/mesh/FT_MB_release_MB_ft_pw3d/
2. Run the following command to infer from the extracted 2D poses:

python infer_wild_mesh.py \
--vid_path <your_video.mp4> \
--json_path <alphapose-results.json> \
--out_path <output_path> \
--ref_3d_motion_path <3d-pose-results.npy> # Optional, use the estimated 3D motion for root trajectory.