Open-Vocabulary Instance Segmentation via Robust Cross-Modal Pseudo-Labeling

TL;DR

This paper introduces a cross-modal pseudo-labeling framework for open-vocabulary instance segmentation, effectively generating training masks by aligning caption semantics with visual features, and employs a robust student model to handle pseudo-mask noise, leading to significant performance improvements.

Contribution

It proposes a novel cross-modal pseudo-labeling approach combined with a noise-aware student model for open-vocabulary segmentation, advancing beyond existing caption-based pretraining methods.

Findings

Improves mAP by 4.5% on MS-COCO

Enhances mAP by 5.1% on Open Images and Conceptual Captions

Demonstrates robustness to pseudo-mask noise

Abstract

Open-vocabulary instance segmentation aims at segmenting novel classes without mask annotations. It is an important step toward reducing laborious human supervision. Most existing works first pretrain a model on captioned images covering many novel classes and then finetune it on limited base classes with mask annotations. However, the high-level textual information learned from caption pretraining alone cannot effectively encode the details required for pixel-wise segmentation. To address this, we propose a cross-modal pseudo-labeling framework, which generates training pseudo masks by aligning word semantics in captions with visual features of object masks in images. Thus, our framework is capable of labeling novel classes in captions via their word semantics to self-train a student model. To account for noises in pseudo masks, we design a robust student model that selectively distills mask knowledge by estimating the mask noise levels, hence mitigating the adverse impact of noisy pseudo masks. By extensive experiments, we show the effectiveness of our framework, where we significantly improve mAP score by 4.5% on MS-COCO and 5.1% on the large-scale Open Images & Conceptual Captions datasets compared to the state-of-the-art.