Enregistré dans:
Détails bibliographiques
Auteurs principaux: Cooper, Avi, Kato, Keizo, Shih, Chia-Hsien, Yamane, Hiroaki, Vinken, Kasper, Takemoto, Kentaro, Sunagawa, Taro, Yeh, Hao-Wei, Yamanaka, Jin, Mason, Ian, Boix, Xavier
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:https://arxiv.org/abs/2410.14690
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908096783187968
author Cooper, Avi
Kato, Keizo
Shih, Chia-Hsien
Yamane, Hiroaki
Vinken, Kasper
Takemoto, Kentaro
Sunagawa, Taro
Yeh, Hao-Wei
Yamanaka, Jin
Mason, Ian
Boix, Xavier
author_facet Cooper, Avi
Kato, Keizo
Shih, Chia-Hsien
Yamane, Hiroaki
Vinken, Kasper
Takemoto, Kentaro
Sunagawa, Taro
Yeh, Hao-Wei
Yamanaka, Jin
Mason, Ian
Boix, Xavier
contents Visual Language Models (VLMs) are now increasingly being merged with Large Language Models (LLMs) to enable new capabilities, particularly in terms of improved interactivity and open-ended responsiveness. While these are remarkable capabilities, the contribution of LLMs to enhancing the longstanding key problem of classifying an image among a set of choices remains unclear. Through extensive experiments involving seven models, ten visual understanding datasets, and multiple prompt variations per dataset, we find that, for object and scene recognition, VLMs that do not leverage LLMs can achieve better performance than VLMs that do. Yet at the same time, leveraging LLMs can improve performance on tasks requiring reasoning and outside knowledge. In response to these challenges, we propose a pragmatic solution: a lightweight fix involving a relatively small LLM that efficiently routes visual tasks to the most suitable model for the task. The LLM router undergoes training using a dataset constructed from more than 2.5 million examples of pairs of visual task and model accuracy. Our results reveal that this lightweight fix surpasses or matches the accuracy of state-of-the-art alternatives, including GPT-4V and HuggingGPT, while improving cost-effectiveness.
format Preprint
id arxiv_https___arxiv_org_abs_2410_14690
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Rethinking VLMs and LLMs for Image Classification
Cooper, Avi
Kato, Keizo
Shih, Chia-Hsien
Yamane, Hiroaki
Vinken, Kasper
Takemoto, Kentaro
Sunagawa, Taro
Yeh, Hao-Wei
Yamanaka, Jin
Mason, Ian
Boix, Xavier
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Visual Language Models (VLMs) are now increasingly being merged with Large Language Models (LLMs) to enable new capabilities, particularly in terms of improved interactivity and open-ended responsiveness. While these are remarkable capabilities, the contribution of LLMs to enhancing the longstanding key problem of classifying an image among a set of choices remains unclear. Through extensive experiments involving seven models, ten visual understanding datasets, and multiple prompt variations per dataset, we find that, for object and scene recognition, VLMs that do not leverage LLMs can achieve better performance than VLMs that do. Yet at the same time, leveraging LLMs can improve performance on tasks requiring reasoning and outside knowledge. In response to these challenges, we propose a pragmatic solution: a lightweight fix involving a relatively small LLM that efficiently routes visual tasks to the most suitable model for the task. The LLM router undergoes training using a dataset constructed from more than 2.5 million examples of pairs of visual task and model accuracy. Our results reveal that this lightweight fix surpasses or matches the accuracy of state-of-the-art alternatives, including GPT-4V and HuggingGPT, while improving cost-effectiveness.
title Rethinking VLMs and LLMs for Image Classification
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.14690