POET: End-to-End Multi-Instance Pose Estimation with Transformers

PyTorch training code and pretrained models for POET (POse Estimation Transformer). We replace the full complex hand-crafted pose estimation pipeline with a Transformer, and outperform Associative Embedding with a ResNet-50, obtaining 54 mAP on COCO. Inference in 50 lines of PyTorch.

For details see End-to-End Trainable Multi-Instance Pose Estimation with Transformers by Lucas Stoffl, Maxime Vidal and Alexander Mathis.

Notebooks

We provide a few notebooks to help you get a grasp on POET:

• POET's notebook: Shows how to load a pre-trained model, generate predictions and visualize the attention of the model.

License

POET is released under the Apache 2.0 license. Please see the LICENSE file for more information.

Contributing

We actively welcome your pull requests! Please see CONTRIBUTING.md and CODE_OF_CONDUCT.md for more info.