On Triangulation as a Form of Self-Supervision for 3D Human Pose Estimation

TL;DR

This paper introduces a semi-supervised method for 3D human pose estimation that leverages multi-view geometric constraints through weighted differentiable triangulation as a form of self-supervision, reducing reliance on labeled data.

Contribution

It proposes a novel triangulation-based self-supervision technique with a weighting mechanism to handle noisy predictions, improving 3D pose estimation in crowded scenes with occlusions.

Findings

Effective on Human3.6M and MPI-INF-3DHP datasets

Handles occlusions and self-occlusion well

Reduces need for extensive labeled data

Abstract

Supervised approaches to 3D pose estimation from single images are remarkably effective when labeled data is abundant. However, as the acquisition of ground-truth 3D labels is labor intensive and time consuming, recent attention has shifted towards semi- and weakly-supervised learning. Generating an effective form of supervision with little annotations still poses major challenge in crowded scenes. In this paper we propose to impose multi-view geometrical constraints by means of a weighted differentiable triangulation and use it as a form of self-supervision when no labels are available. We therefore train a 2D pose estimator in such a way that its predictions correspond to the re-projection of the triangulated 3D pose and train an auxiliary network on them to produce the final 3D poses. We complement the triangulation with a weighting mechanism that alleviates the impact of noisy predictions caused by self-occlusion or occlusion from other subjects. We demonstrate the effectiveness of our semi-supervised approach on Human3.6M and MPI-INF-3DHP datasets, as well as on a new multi-view multi-person dataset that features occlusion.