CVMar 27, 2022

Locality-Aware Inter-and Intra-Video Reconstruction for Self-Supervised Correspondence Learning

arXiv:2203.14333v258 citationsh-index: 77
Originality Incremental advance
AI Analysis

This work addresses the problem of self-supervised correspondence learning for computer vision researchers, offering incremental improvements by combining existing ideas into a unified framework.

The paper tackled learning visual correspondence from unlabeled videos by developing LIIR, a framework that integrates instance discrimination, location awareness, and spatial compactness, resulting in a learned representation that surpasses self-supervised state-of-the-art methods on label propagation tasks for objects, semantic parts, and keypoints.

Our target is to learn visual correspondence from unlabeled videos. We develop LIIR, a locality-aware inter-and intra-video reconstruction framework that fills in three missing pieces, i.e., instance discrimination, location awareness, and spatial compactness, of self-supervised correspondence learning puzzle. First, instead of most existing efforts focusing on intra-video self-supervision only, we exploit cross video affinities as extra negative samples within a unified, inter-and intra-video reconstruction scheme. This enables instance discriminative representation learning by contrasting desired intra-video pixel association against negative inter-video correspondence. Second, we merge position information into correspondence matching, and design a position shifting strategy to remove the side-effect of position encoding during inter-video affinity computation, making our LIIR location-sensitive. Third, to make full use of the spatial continuity nature of video data, we impose a compactness-based constraint on correspondence matching, yielding more sparse and reliable solutions. The learned representation surpasses self-supervised state-of-the-arts on label propagation tasks including objects, semantic parts, and keypoints.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes