Region-Based Representations Revisited

TL;DR

This paper demonstrates that combining recent class-agnostic segmenters with strong unsupervised features enables effective region-based recognition across various tasks, offering advantages in flexibility and efficiency.

Contribution

It revisits region-based representations by integrating SAM and DINOv2, showing their effectiveness for diverse recognition tasks with simple linear decoders.

Findings

Effective for semantic segmentation and image retrieval

Competitive performance with simple linear decoders

Suitable for video analysis and large-scale inference

Abstract

We investigate whether region-based representations are effective for recognition. Regions were once a mainstay in recognition approaches, but pixel and patch-based features are now used almost exclusively. We show that recent class-agnostic segmenters like SAM can be effectively combined with strong unsupervised representations like DINOv2 and used for a wide variety of tasks, including semantic segmentation, object-based image retrieval, and multi-image analysis. Once the masks and features are extracted, these representations, even with linear decoders, enable competitive performance, making them well suited to applications that require custom queries. The compactness of the representation also makes it well-suited to video analysis and other problems requiring inference across many images.