Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Dense and Aligned Captions (DAC) Promote Compositional Reasoning in VL Models

About

Vision and Language (VL) models offer an effective method for aligning representation spaces of images and text, leading to numerous applications such as cross-modal retrieval, visual question answering, captioning, and more. However, the aligned image-text spaces learned by all the popular VL models are still suffering from the so-called `object bias' - their representations behave as `bags of nouns', mostly ignoring or downsizing the attributes, relations, and states of objects described/appearing in texts/images. Although some great attempts at fixing these `compositional reasoning' issues were proposed in the recent literature, the problem is still far from being solved. In this paper, we uncover two factors limiting the VL models' compositional reasoning performance. These two factors are properties of the paired VL dataset used for finetuning and pre-training the VL model: (i) the caption quality, or in other words `image-alignment', of the texts; and (ii) the `density' of the captions in the sense of mentioning all the details appearing on the image. We propose a fine-tuning approach for automatically treating these factors leveraging a standard VL dataset (CC3M). Applied to CLIP, we demonstrate its significant compositional reasoning performance increase of up to $\sim27\%$ over the base model, up to $\sim20\%$ over the strongest baseline, and by $6.7\%$ on average.

Sivan Doveh, Assaf Arbelle, Sivan Harary, Roei Herzig, Donghyun Kim, Paola Cascante-bonilla, Amit Alfassy, Rameswar Panda, Raja Giryes, Rogerio Feris, Shimon Ullman, Leonid Karlinsky• 2023

Related benchmarks

TaskDatasetResultRank
Image ClassificationImageNet-1K
Top-1 Acc51.02
600
Text-to-Image RetrievalFlickr30K
R@126.19
559
Text-to-Image RetrievalFlickr30k (test)
Recall@164.9
525
Image-to-Text RetrievalFlickr30k (test)
R@153.1
472
Image-to-Text RetrievalFlickr30K
R@122.71
451
Image ClassificationCIFAR100
Accuracy64.1
301
Image ClassificationCIFAR10
Accuracy (%)90.4
282
Image RetrievalMS-COCO--
172
Object DetectionCOCO
mAP24.74
137
Text RetrievalFlickr30K--
120
Showing 10 of 46 rows

Other info

Follow for update