Self-supervised Learning for Video Correspondence Flow

TL;DR

This paper introduces a self-supervised learning approach for video correspondence flow, leveraging natural video coherence to learn robust features for matching, and achieves state-of-the-art results on key video tasks.

Contribution

It proposes a novel self-supervised framework with an information bottleneck, recursive training over long sequences, and demonstrates significant performance improvements.

Findings

State-of-the-art results on DAVIS 2017 and JHMDB datasets.

Robust feature learning without manual annotations.

Enhanced performance with additional diverse training data.

Abstract

The objective of this paper is self-supervised learning of feature embeddings that are suitable for matching correspondences along the videos, which we term correspondence flow. By leveraging the natural spatial-temporal coherence in videos, we propose to train a ``pointer'' that reconstructs a target frame by copying pixels from a reference frame. We make the following contributions: First, we introduce a simple information bottleneck that forces the model to learn robust features for correspondence matching, and prevent it from learning trivial solutions, \eg matching based on low-level colour information. Second, to tackle the challenges from tracker drifting, due to complex object deformations, illumination changes and occlusions, we propose to train a recursive model over long temporal windows with scheduled sampling and cycle consistency. Third, we achieve state-of-the-art performance on DAVIS 2017 video segmentation and JHMDB keypoint tracking tasks, outperforming all previous self-supervised learning approaches by a significant margin. Fourth, in order to shed light on the potential of self-supervised learning on the task of video correspondence flow, we probe the upper bound by training on additional data, \ie more diverse videos, further demonstrating significant improvements on video segmentation.