Linear Alignment of Vision-language Models for Image Captioning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Paischer, Fabian, Hofmarcher, Markus, Hochreiter, Sepp, Adler, Thomas
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916604977086464
author Paischer, Fabian
Hofmarcher, Markus
Hochreiter, Sepp
Adler, Thomas
author_facet Paischer, Fabian
Hofmarcher, Markus
Hochreiter, Sepp
Adler, Thomas
contents Recently, vision-language models like CLIP have advanced the state of the art in a variety of multi-modal tasks including image captioning and caption evaluation. Many approaches leverage CLIP for cross-modal retrieval to condition pre-trained language models on visual input. However, CLIP generally suffers from a mis-alignment of image and text modalities in the joint embedding space. We investigate efficient methods to linearly re-align the joint embedding space for the downstream task of image captioning. This leads to an efficient training protocol that merely requires computing a closed-form solution for a linear mapping in the joint CLIP space. Consequently, we propose a lightweight captioning method called ReCap, which can be trained up to 1000 times faster than existing lightweight methods. Moreover, we propose two new learning-based image-captioning metrics built on CLIP score along with our proposed alignment. We evaluate ReCap on MS-COCO, Flickr30k, VizWiz and MSRVTT. On the former two, ReCap performs comparably to state-of-the-art lightweight methods using rule-based metrics while outperforming them on most of the CLIP-based metrics. On the latter two benchmarks, ReCap consistently outperforms competitors across all metrics and exhibits strong transfer capabilities and resilience to noise. Finally, we demonstrate that our proposed metrics correlate stronger with human judgement than existing metrics on the Flickr8k-Expert, Flickr8k-Crowdflower, and THumB datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2307_05591
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Linear Alignment of Vision-language Models for Image Captioning
Paischer, Fabian
Hofmarcher, Markus
Hochreiter, Sepp
Adler, Thomas
Computer Vision and Pattern Recognition
Computation and Language
Machine Learning
Recently, vision-language models like CLIP have advanced the state of the art in a variety of multi-modal tasks including image captioning and caption evaluation. Many approaches leverage CLIP for cross-modal retrieval to condition pre-trained language models on visual input. However, CLIP generally suffers from a mis-alignment of image and text modalities in the joint embedding space. We investigate efficient methods to linearly re-align the joint embedding space for the downstream task of image captioning. This leads to an efficient training protocol that merely requires computing a closed-form solution for a linear mapping in the joint CLIP space. Consequently, we propose a lightweight captioning method called ReCap, which can be trained up to 1000 times faster than existing lightweight methods. Moreover, we propose two new learning-based image-captioning metrics built on CLIP score along with our proposed alignment. We evaluate ReCap on MS-COCO, Flickr30k, VizWiz and MSRVTT. On the former two, ReCap performs comparably to state-of-the-art lightweight methods using rule-based metrics while outperforming them on most of the CLIP-based metrics. On the latter two benchmarks, ReCap consistently outperforms competitors across all metrics and exhibits strong transfer capabilities and resilience to noise. Finally, we demonstrate that our proposed metrics correlate stronger with human judgement than existing metrics on the Flickr8k-Expert, Flickr8k-Crowdflower, and THumB datasets.
title Linear Alignment of Vision-language Models for Image Captioning
topic Computer Vision and Pattern Recognition
Computation and Language
Machine Learning
url https://arxiv.org/abs/2307.05591