3D human pose estimation with adaptive receptive fields and dilated temporal convolutions

TL;DR

This paper introduces adaptive receptive fields for 3D human pose estimation using optical flow, enabling faster processing of slow-motion sequences with minimal accuracy loss.

Contribution

The paper proposes a novel adaptive receptive field method that improves efficiency and speed in 3D pose estimation models based on optical flow.

Findings

Model with adaptive receptive fields is 23% faster on slow-motion sequences.

Achieves similar accuracy to benchmark with reduced receptive fields.

Enables processing of 10x longer sequences without significant accuracy loss.

Abstract

In this work, we demonstrate that receptive fields in 3D pose estimation can be effectively specified using optical flow. We introduce adaptive receptive fields, a simple and effective method to aid receptive field selection in pose estimation models based on optical flow inference. We contrast the performance of a benchmark state-of-the-art model running on fixed receptive fields with their adaptive field counterparts. By using a reduced receptive field, our model can process slow-motion sequences (10x longer) 23% faster than the benchmark model running at regular speed. The reduction in computational cost is achieved while producing a pose prediction accuracy to within 0.36% of the benchmark model.