Joint-task Self-supervised Learning for Temporal Correspondence

TL;DR

This paper introduces a self-supervised learning approach that jointly learns dense temporal correspondence in videos by integrating region tracking and pixel-level matching through a shared affinity matrix, outperforming existing methods.

Contribution

It presents a novel joint-task framework that leverages the synergy between region and pixel-level tasks for improved video correspondence learning.

Findings

Outperforms state-of-the-art self-supervised methods on multiple tasks

Surpasses fully-supervised ResNet-18 features in affinity representation

Effective in video-object, part-segmentation, keypoint, and object tracking

Abstract

This paper proposes to learn reliable dense correspondence from videos in a self-supervised manner. Our learning process integrates two highly related tasks: tracking large image regions \emph{and} establishing fine-grained pixel-level associations between consecutive video frames. We exploit the synergy between both tasks through a shared inter-frame affinity matrix, which simultaneously models transitions between video frames at both the region- and pixel-levels. While region-level localization helps reduce ambiguities in fine-grained matching by narrowing down search regions; fine-grained matching provides bottom-up features to facilitate region-level localization. Our method outperforms the state-of-the-art self-supervised methods on a variety of visual correspondence tasks, including video-object and part-segmentation propagation, keypoint tracking, and object tracking. Our self-supervised method even surpasses the fully-supervised affinity feature representation obtained from a ResNet-18 pre-trained on the ImageNet.