ReCo: Retrieve and Co-segment for Zero-shot Transfer

TL;DR

ReCo combines language-image pre-training and retrieval to enable zero-shot, nameable segmentation without pixel labels, effectively segmenting rare objects and reducing annotation costs.

Contribution

It introduces a novel method that synthesizes retrieval and co-segmentation to achieve zero-shot, nameable segmentation leveraging CLIP's capabilities.

Findings

ReCo outperforms existing unsupervised segmentation methods.

It enables zero-shot transfer with nameable predictions.

ReCo can generate segmenters for rare objects.

Abstract

Semantic segmentation has a broad range of applications, but its real-world impact has been significantly limited by the prohibitive annotation costs necessary to enable deployment. Segmentation methods that forgo supervision can side-step these costs, but exhibit the inconvenient requirement to provide labelled examples from the target distribution to assign concept names to predictions. An alternative line of work in language-image pre-training has recently demonstrated the potential to produce models that can both assign names across large vocabularies of concepts and enable zero-shot transfer for classification, but do not demonstrate commensurate segmentation abilities. In this work, we strive to achieve a synthesis of these two approaches that combines their strengths. We leverage the retrieval abilities of one such language-image pre-trained model, CLIP, to dynamically curate training sets from unlabelled images for arbitrary collections of concept names, and leverage the robust correspondences offered by modern image representations to co-segment entities among the resulting collections. The synthetic segment collections are then employed to construct a segmentation model (without requiring pixel labels) whose knowledge of concepts is inherited from the scalable pre-training process of CLIP. We demonstrate that our approach, termed Retrieve and Co-segment (ReCo) performs favourably to unsupervised segmentation approaches while inheriting the convenience of nameable predictions and zero-shot transfer. We also demonstrate ReCo's ability to generate specialist segmenters for extremely rare objects.