ArtVLM: Attribute Recognition Through Vision-Based Prefix Language Modeling

TL;DR

ArtVLM introduces a novel sentence generation-based retrieval method for zero-shot visual attribute recognition, explicitly modeling object-attribute relations as a dependency-sensitive language modeling task, outperforming contrastive methods.

Contribution

The paper proposes a new generative retrieval approach that captures object-attribute dependencies for improved zero-shot attribute recognition using vision-language models.

Findings

Generative retrieval outperforms contrastive retrieval on VAW and VGARank datasets.

Explicit modeling of object-attribute relations improves recognition accuracy.

Method effectively distills knowledge from large pretrained vision-language models.

Abstract

Recognizing and disentangling visual attributes from objects is a foundation to many computer vision applications. While large vision language representations like CLIP had largely resolved the task of zero-shot object recognition, zero-shot visual attribute recognition remains a challenge because CLIP's contrastively-learned vision-language representation cannot effectively capture object-attribute dependencies. In this paper, we target this weakness and propose a sentence generation-based retrieval formulation for attribute recognition that is novel in 1) explicitly modeling a to-be-measured and retrieved object-attribute relation as a conditional probability graph, which converts the recognition problem into a dependency-sensitive language-modeling problem, and 2) applying a large pretrained Vision-Language Model (VLM) on this reformulation and naturally distilling its knowledge of image-object-attribute relations to use towards attribute recognition. Specifically, for each attribute to be recognized on an image, we measure the visual-conditioned probability of generating a short sentence encoding the attribute's relation to objects on the image. Unlike contrastive retrieval, which measures likelihood by globally aligning elements of the sentence to the image, generative retrieval is sensitive to the order and dependency of objects and attributes in the sentence. We demonstrate through experiments that generative retrieval consistently outperforms contrastive retrieval on two visual reasoning datasets, Visual Attribute in the Wild (VAW), and our newly-proposed Visual Genome Attribute Ranking (VGARank).