HITrack 1.0.4

3D scene on a monocular video

HITrack

HITrack or Human Inertial Tracking is a pipeline consisting of 3 human recognition state-of-the-art neural networks (yolov7, VitPose and MHFormer) linked together by specially designed Inertial Tracking to produce a 3D scene on a monocular image.

Quick start

pip install HITrack

from HITrack import HITrack
hit = HITrack('videos/dance.mp4')

# 2D keypoints + tracking
hit.compute_2d(yolo='yolov7x', vitpose='b')

# merging recovered tracks and broken tracks manually
hit.recover_2d({2:4, 3:5})

# 2D to 3D by tracking
hit.compute_3d()

# 3D to scene
hit.compute_scene()

# visualising any of these steps
hit.visualize('3D_scene', compress=True, original_sound=True)

License

This project is licensed under the terms of the MIT license.