CoDet: Co-Occurrence Guided Region-Word Alignment for Open-Vocabulary Object Detection

TL;DR

CoDet introduces a co-occurrence based method for aligning image regions with words, improving open-vocabulary object detection by discovering shared objects through visual similarities and co-occurrence in image groups.

Contribution

The paper presents CoDet, a novel approach that reformulates region-word alignment as a co-occurring object discovery problem, avoiding reliance on pre-aligned vision-language spaces.

Findings

Achieves 37.0 AP^m_{novel} on OV-LVIS, surpassing previous SOTA by 4.2 points.

Demonstrates superior scalability and performance in open-vocabulary detection.

Effectively discovers shared objects through co-occurrence and visual similarity.

Abstract

Deriving reliable region-word alignment from image-text pairs is critical to learn object-level vision-language representations for open-vocabulary object detection. Existing methods typically rely on pre-trained or self-trained vision-language models for alignment, which are prone to limitations in localization accuracy or generalization capabilities. In this paper, we propose CoDet, a novel approach that overcomes the reliance on pre-aligned vision-language space by reformulating region-word alignment as a co-occurring object discovery problem. Intuitively, by grouping images that mention a shared concept in their captions, objects corresponding to the shared concept shall exhibit high co-occurrence among the group. CoDet then leverages visual similarities to discover the co-occurring objects and align them with the shared concept. Extensive experiments demonstrate that CoDet has superior performances and compelling scalability in open-vocabulary detection, e.g., by scaling up the visual backbone, CoDet achieves 37.0 $\text{AP}^m_{novel}$ and 44.7 $\text{AP}^m_{all}$ on OV-LVIS, surpassing the previous SoTA by 4.2 $\text{AP}^m_{novel}$ and 9.8 $\text{AP}^m_{all}$. Code is available at https://github.com/CVMI-Lab/CoDet.