Action Transformer: A Self-Attention Model for Short-Time Pose-Based Human Action Recognition

TL;DR

The paper introduces Action Transformer, a fully self-attentional model for real-time human action recognition using 2D pose data, outperforming complex architectures and establishing a new benchmark dataset.

Contribution

It presents a novel self-attentional architecture for HAR and introduces MPOSE2021, a large-scale dataset for benchmarking real-time action recognition.

Findings

Action Transformer outperforms existing models on MPOSE2021.

The approach achieves low latency and high accuracy in real-time HAR.

The dataset facilitates standardized evaluation for short-time HAR.

Abstract

Deep neural networks based purely on attention have been successful across several domains, relying on minimal architectural priors from the designer. In Human Action Recognition (HAR), attention mechanisms have been primarily adopted on top of standard convolutional or recurrent layers, improving the overall generalization capability. In this work, we introduce Action Transformer (AcT), a simple, fully self-attentional architecture that consistently outperforms more elaborated networks that mix convolutional, recurrent and attentive layers. In order to limit computational and energy requests, building on previous human action recognition research, the proposed approach exploits 2D pose representations over small temporal windows, providing a low latency solution for accurate and effective real-time performance. Moreover, we open-source MPOSE2021, a new large-scale dataset, as an attempt to build a formal training and evaluation benchmark for real-time, short-time HAR. The proposed methodology was extensively tested on MPOSE2021 and compared to several state-of-the-art architectures, proving the effectiveness of the AcT model and laying the foundations for future work on HAR.