Bootstrapping Human Optical Flow and Pose

TL;DR

This paper introduces a bootstrapping framework that jointly improves human optical flow and pose estimation by mutual refinement, achieving state-of-the-art results on multiple datasets.

Contribution

It presents a novel joint optimization approach that enhances both tasks simultaneously by leveraging their interdependence during training.

Findings

Achieves state-of-the-art pose estimation accuracy.

Improves optical flow accuracy at human joints.

Demonstrates benefits of joint task optimization.

Abstract

We propose a bootstrapping framework to enhance human optical flow and pose. We show that, for videos involving humans in scenes, we can improve both the optical flow and the pose estimation quality of humans by considering the two tasks at the same time. We enhance optical flow estimates by fine-tuning them to fit the human pose estimates and vice versa. In more detail, we optimize the pose and optical flow networks to, at inference time, agree with each other. We show that this results in state-of-the-art results on the Human 3.6M and 3D Poses in the Wild datasets, as well as a human-related subset of the Sintel dataset, both in terms of pose estimation accuracy and the optical flow accuracy at human joint locations. Code available at https://github.com/ubc-vision/bootstrapping-human-optical-flow-and-pose