ViLLA: Fine-Grained Vision-Language Representation Learning from Real-World Data

TL;DR

This paper introduces ViLLA, a novel approach for training vision-language models on complex, real-world data with fine-grained region-attribute relationships, significantly improving fine-grained reasoning tasks.

Contribution

ViLLA is a new framework that decomposes complex image-text pairs into region-attribute pairs and learns from these, addressing the limitations of standard VLMs on complex datasets.

Findings

ViLLA outperforms existing VLMs on fine-grained reasoning tasks.

Standard VLMs struggle with high pairwise complexity, showing up to 37% performance degradation.

ViLLA achieves up to 3.6 AP50 points in zero-shot object detection and 14.2 R-Precision in retrieval.

Abstract

Vision-language models (VLMs), such as CLIP and ALIGN, are generally trained on datasets consisting of image-caption pairs obtained from the web. However, real-world multimodal datasets, such as healthcare data, are significantly more complex: each image (e.g. X-ray) is often paired with text (e.g. physician report) that describes many distinct attributes occurring in fine-grained regions of the image. We refer to these samples as exhibiting high pairwise complexity, since each image-text pair can be decomposed into a large number of region-attribute pairings. The extent to which VLMs can capture fine-grained relationships between image regions and textual attributes when trained on such data has not been previously evaluated. The first key contribution of this work is to demonstrate through systematic evaluations that as the pairwise complexity of the training dataset increases, standard VLMs struggle to learn region-attribute relationships, exhibiting performance degradations of up to 37% on retrieval tasks. In order to address this issue, we introduce ViLLA as our second key contribution. ViLLA, which is trained to capture fine-grained region-attribute relationships from complex datasets, involves two components: (a) a lightweight, self-supervised mapping model to decompose image-text samples into region-attribute pairs, and (b) a contrastive VLM to learn representations from generated region-attribute pairs. We demonstrate with experiments across four domains (synthetic, product, medical, and natural images) that ViLLA outperforms comparable VLMs on fine-grained reasoning tasks, such as zero-shot object detection (up to 3.6 AP50 points on COCO and 0.6 mAP points on LVIS) and retrieval (up to 14.2 R-Precision points).