Summary of Rewind: Understanding Long Videos with Instructed Learnable Memory, by Anxhelo Diko et al.
ReWind: Understanding Long Videos with Instructed Learnable Memory
by Anxhelo Diko, Tinghuai Wang, Wassim Swaileh, Shiyan Sun, Ioannis Patras
First submitted to arxiv on: 23 Nov 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
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 paper introduces ReWind, a novel vision-language model designed to efficiently understand long videos while maintaining coherent understanding across extended sequences. The two-stage framework consists of a dynamic learnable memory module that stores and updates instruction-relevant visual information as the video unfolds, and an adaptive frame selection mechanism guided by the memory content to identify key moments. This approach allows for low memory requirements and scaling linearly with the number of tokens. ReWind is evaluated on visual question answering (VQA) and temporal grounding tasks, demonstrating superior performance on long video benchmarks. The model achieves a +13% score gain and +12% accuracy improvement on MovieChat-1K VQA dataset, as well as an +8% mIoU increase on Charades-STA for temporal grounding. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary ReWind is a new way to understand videos that are very long. Right now, computers struggle to keep up with these long videos because they need too much memory and it takes too long to process them. ReWind is different because it uses a special memory system that can learn as the video plays. This helps the computer focus on the most important parts of the video and ignore the rest. The result is a better understanding of what’s happening in the video, which is useful for tasks like answering questions about what you’re seeing. |
Keywords
» Artificial intelligence » Grounding » Language model » Question answering
