Summary of Kangaroo: a Powerful Video-language Model Supporting Long-context Video Input, by Jiajun Liu et al.
Kangaroo: A Powerful Video-Language Model Supporting Long-context Video Input
by Jiajun Liu, Yibing Wang, Hanghang Ma, Xiaoping Wu, Xiaoqi Ma, Xiaoming Wei, Jianbin Jiao, Enhua Wu, Jie Hu
First submitted to arxiv on: 28 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Multimedia (cs.MM)
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 abstract presents a paper that introduces a new Large Multi-modal Model (LMM) called Kangaroo, designed to address the challenges of processing long videos. Existing methods struggle due to limited access to high-quality video data and excessive compression of visual features. The authors develop a data curation system to build a large-scale dataset with annotations for vision-language pre-training and instruction tuning. They also design a curriculum training pipeline that gradually increases resolution and input frames to accommodate long videos. Kangaroo achieves state-of-the-art performance on various video understanding benchmarks, outperforming larger models and proprietary ones on tasks specific to long videos. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Kangaroo is a new model designed to understand long videos better. Currently, it’s hard to make computers understand long videos because we don’t have enough good data and the visual features are compressed too much. The people who wrote this paper created a system to collect a lot of video data with good annotations for training their model. They also made a special way to train the model so it can handle longer videos. Their model, Kangaroo, is really good at understanding long videos and does better than some other models on certain tasks. |
Keywords
» Artificial intelligence » Instruction tuning » Multi modal
