Home

Awesome

3D Human Pose Estimation with Spatial and Temporal Transformers

This repo is the official implementation for CrossFormer: Cross Spatio-Temporal Transformer for 3D Human Pose Estimation

Our code is built on top of VideoPose3D.

Environment

The code is developed and tested under the following environment

You can create the environment:

conda env create -f crossformer.yml

Dataset

Our code is compatible with the dataset setup introduced by Martinez et al. and Pavllo et al.. Please refer to VideoPose3D to set up the Human3.6M dataset (./data directory).

Evaluating pre-trained models

We provide the pre-trained 81-frame model (CPN detected 2D pose as input) here. To evaluate it, put it into the ./checkpoint directory and run:

python run_crossformer.py -k cpn_ft_h36m_dbb -f 81 -c checkpoint --evaluate best_epoch44.4.bin

We also provide pre-trained 81-frame model (Ground truth 2D pose as input) here. To evaluate it, put it into the ./checkpoint directory and run:

python run_crossformer.py -k gt -f 81 -c checkpoint --evaluate best_epoch_gt_28.5.bin

Training new models

python run_crossformer.py -k cpn_ft_h36m_dbb -f 27 -lr 0.00004 -lrd 0.99

python run_crossformer.py -k gt -f 81 -lr 0.0004 -lrd 0.99

81 frames achieves 28.5 mm (MPJPE).

Visualization and other functions

We keep our code consistent with VideoPose3D. Please refer to their project page for further information.

Acknowledgement

Part of our code is borrowed from VideoPose3D. We thank the authors for releasing the codes.