STAR-Pose: Efficient Low-Resolution Video Human Pose Estimation via Spatial-Temporal Adaptive Super-Resolution

TL;DR

STAR-Pose introduces a novel spatial-temporal super-resolution framework with a Transformer and adaptive fusion, significantly improving low-resolution video human pose estimation efficiency and accuracy.

Contribution

It presents a new adaptive super-resolution method with a specialized Transformer and pose-aware loss for better keypoint localization in low-res videos.

Findings

Achieves up to 5.2% mAP improvement at 64x48 resolution.

Faster inference by 2.8x to 4.4x compared to cascaded methods.

Outperforms existing approaches on multiple datasets.

Abstract

Human pose estimation in low-resolution videos presents a fundamental challenge in computer vision. Conventional methods either assume high-quality inputs or employ computationally expensive cascaded processing, which limits their deployment in resource-constrained environments. We propose STAR-Pose, a spatial-temporal adaptive super-resolution framework specifically designed for video-based human pose estimation. Our method features a novel spatial-temporal Transformer with LeakyReLU-modified linear attention, which efficiently captures long-range temporal dependencies. Moreover, it is complemented by an adaptive fusion module that integrates parallel CNN branch for local texture enhancement. We also design a pose-aware compound loss to achieve task-oriented super-resolution. This loss guides the network to reconstruct structural features that are most beneficial for keypoint localization, rather than optimizing purely for visual quality. Extensive experiments on several mainstream video HPE datasets demonstrate that STAR-Pose outperforms existing approaches. It achieves up to 5.2% mAP improvement under extremely low-resolution (64x48) conditions while delivering 2.8x to 4.4x faster inference than cascaded approaches.