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

VinVL: Revisiting Visual Representations in Vision-Language Models

About

This paper presents a detailed study of improving visual representations for vision language (VL) tasks and develops an improved object detection model to provide object-centric representations of images. Compared to the most widely used \emph{bottom-up and top-down} model \cite{anderson2018bottom}, the new model is bigger, better-designed for VL tasks, and pre-trained on much larger training corpora that combine multiple public annotated object detection datasets. Therefore, it can generate representations of a richer collection of visual objects and concepts. While previous VL research focuses mainly on improving the vision-language fusion model and leaves the object detection model improvement untouched, we show that visual features matter significantly in VL models. In our experiments we feed the visual features generated by the new object detection model into a Transformer-based VL fusion model \oscar \cite{li2020oscar}, and utilize an improved approach \short\ to pre-train the VL model and fine-tune it on a wide range of downstream VL tasks. Our results show that the new visual features significantly improve the performance across all VL tasks, creating new state-of-the-art results on seven public benchmarks. We will release the new object detection model to public.

Pengchuan Zhang, Xiujun Li, Xiaowei Hu, Jianwei Yang, Lei Zhang, Lijuan Wang, Yejin Choi, Jianfeng Gao• 2021

Related benchmarks

TaskDatasetResultRank
Visual Question AnsweringVQA v2 (test-dev)
Overall Accuracy76.6
706
Image CaptioningMS COCO Karpathy (test)
CIDEr140.4
682
Visual Question AnsweringVQA v2 (test-std)
Accuracy76.6
486
Referring Expression ComprehensionRefCOCO+ (val)--
354
Natural Language Visual ReasoningNLVR2 (test-p)
Accuracy84
346
Referring Expression ComprehensionRefCOCO (val)--
344
Referring Expression ComprehensionRefCOCO (testA)--
342
Visual Question AnsweringVQA 2.0 (test-dev)
Accuracy76.52
337
Image-to-Text RetrievalMS-COCO 5K (test)
R@175.4
320
Text-to-Image RetrievalMSCOCO 5K (test)
R@175.4
308
Showing 10 of 72 rows
...

Other info

Code

Follow for update