Summary of Unified Video-language Pre-training with Synchronized Audio, by Shentong Mo et al.
Unified Video-Language Pre-training with Synchronized Audio
by Shentong Mo, Haofan Wang, Huaxia Li, Xu Tang
First submitted to arxiv on: 12 May 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM); Sound (cs.SD); Audio and Speech Processing (eess.AS)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed Video-Language pre-training framework, VLSA, learns tri-modal representations from large-scale data in a self-supervised way. It uses a unified transformer to jointly process video, text, and audio modalities. The model incorporates local-patch masked modeling to learn modality-aware features and global audio matching to capture audio-guided features for video and text. VLSA outperforms state-of-the-art baselines on retrieval tasks across text, video, and audio, even with limited training data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary VLSA is a new way to train machines to understand videos, texts, and sounds together. It’s like teaching a computer to watch a movie, read the subtitles, and listen to the soundtrack at the same time! The researchers created this system to make it better at recognizing patterns in all three types of media. They tested it on lots of data and found that it was really good at matching videos with texts and sounds. |
Keywords
» Artificial intelligence » Self supervised » Transformer
