Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

TVLT: Textless Vision-Language Transformer

About

In this work, we present the Textless Vision-Language Transformer (TVLT), where homogeneous transformer blocks take raw visual and audio inputs for vision-and-language representation learning with minimal modality-specific design, and do not use text-specific modules such as tokenization or automatic speech recognition (ASR). TVLT is trained by reconstructing masked patches of continuous video frames and audio spectrograms (masked autoencoding) and contrastive modeling to align video and audio. TVLT attains performance comparable to its text-based counterpart on various multimodal tasks, such as visual question answering, image retrieval, video retrieval, and multimodal sentiment analysis, with 28x faster inference speed and only 1/3 of the parameters. Our findings suggest the possibility of learning compact and efficient visual-linguistic representations from low-level visual and audio signals without assuming the prior existence of text. Our code and checkpoints are available at: https://github.com/zinengtang/TVLT

Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal• 2022

Related benchmarks

TaskDatasetResultRank
Visual Question AnsweringVQA v2
Accuracy61
1165
Text-to-Video RetrievalMSR-VTT
Recall@122
313
Audio-to-Visual RetrievalMSR-VTT (test)
R@122.6
18
Audio-to-Video RetrievalMSR-VTT
Recall@123.8
13
Visual Question AnsweringVQA v1
Accuracy58.9
4
Audio-to-image retrievalPlaces 400k
R@149
4
Audio-to-Video RetrievalYouCook (test)
Recall@131.8
4
Audio-to-Video RetrievalYouCook2
Recall@132.8
3
Audio-to-Video RetrievalCrossTask
Recall@115.3
3
Multimodal Sentiment AnalysisCMU-MOSEI
A2 Score76.8
3
Showing 10 of 10 rows

Other info

Code

Follow for update