Human Pose Regression with Residual Log-likelihood Estimation

TL;DR

This paper introduces Residual Log-likelihood Estimation (RLE), a novel regression-based approach for human pose estimation that models output distribution more accurately, outperforming traditional heatmap methods in accuracy and efficiency.

Contribution

The paper proposes RLE, a new regression paradigm that captures output distribution changes, compatible with flow models, improving pose estimation performance significantly.

Findings

12.4 mAP improvement on MSCOCO without test-time overhead

Outperforms heatmap-based methods in multi-person pose estimation

Effective and flexible regression method for human pose tasks

Abstract

Heatmap-based methods dominate in the field of human pose estimation by modelling the output distribution through likelihood heatmaps. In contrast, regression-based methods are more efficient but suffer from inferior performance. In this work, we explore maximum likelihood estimation (MLE) to develop an efficient and effective regression-based methods. From the perspective of MLE, adopting different regression losses is making different assumptions about the output density function. A density function closer to the true distribution leads to a better regression performance. In light of this, we propose a novel regression paradigm with Residual Log-likelihood Estimation (RLE) to capture the underlying output distribution. Concretely, RLE learns the change of the distribution instead of the unreferenced underlying distribution to facilitate the training process. With the proposed reparameterization design, our method is compatible with off-the-shelf flow models. The proposed method is effective, efficient and flexible. We show its potential in various human pose estimation tasks with comprehensive experiments. Compared to the conventional regression paradigm, regression with RLE bring 12.4 mAP improvement on MSCOCO without any test-time overhead. Moreover, for the first time, especially on multi-person pose estimation, our regression method is superior to the heatmap-based methods. Our code is available at https://github.com/Jeff-sjtu/res-loglikelihood-regression