TransPose: Keypoint Localization via Transformer

TL;DR

TransPose introduces a Transformer-based model for human pose estimation that captures long-range dependencies, improves localization accuracy, and provides interpretability, outperforming CNN-based methods on COCO and MPII benchmarks.

Contribution

The paper presents TransPose, a novel Transformer-based architecture for pose estimation that captures spatial dependencies and offers interpretability, outperforming CNN models in accuracy and efficiency.

Findings

Achieves 75.8 AP on COCO validation set.

Outperforms CNN architectures in speed and accuracy.

Transfers well to MPII benchmark with minimal training.

Abstract

While CNN-based models have made remarkable progress on human pose estimation, what spatial dependencies they capture to localize keypoints remains unclear. In this work, we propose a model called \textbf{TransPose}, which introduces Transformer for human pose estimation. The attention layers built in Transformer enable our model to capture long-range relationships efficiently and also can reveal what dependencies the predicted keypoints rely on. To predict keypoint heatmaps, the last attention layer acts as an aggregator, which collects contributions from image clues and forms maximum positions of keypoints. Such a heatmap-based localization approach via Transformer conforms to the principle of Activation Maximization~\cite{erhan2009visualizing}. And the revealed dependencies are image-specific and fine-grained, which also can provide evidence of how the model handles special cases, e.g., occlusion. The experiments show that TransPose achieves 75.8 AP and 75.0 AP on COCO validation and test-dev sets, while being more lightweight and faster than mainstream CNN architectures. The TransPose model also transfers very well on MPII benchmark, achieving superior performance on the test set when fine-tuned with small training costs. Code and pre-trained models are publicly available\footnote{\url{https://github.com/yangsenius/TransPose}}.