Summary of Froster: Frozen Clip Is a Strong Teacher For Open-vocabulary Action Recognition, by Xiaohu Huang et al.
FROSTER: Frozen CLIP Is A Strong Teacher for Open-Vocabulary Action Recognition
by Xiaohu Huang, Hao Zhou, Kun Yao, Kai Han
First submitted to arxiv on: 5 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 FROSTER, a framework for open-vocabulary action recognition, leverages the strong generalization capability of the CLIP model. While CLIP has achieved success in various image-based tasks due to its massive pretraining on image-text pairs, applying it directly to open-vocabulary action recognition is challenging due to the lack of temporal information in CLIP’s pretraining. Furthermore, fine-tuning CLIP on action recognition datasets may lead to overfitting and hinder generalizability, resulting in unsatisfactory results when dealing with unseen actions. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary FROSTER helps computers recognize actions without a predefined list of actions. It uses the strong learning ability of a model called CLIP, which has learned to understand images and text by looking at many image-text pairs. This ability helps CLIP recognize things in images well. However, this same ability makes it hard for CLIP to recognize actions because it doesn’t know about time. When we try to teach CLIP to recognize actions, it can get too good at recognizing the specific actions it learned and not be able to recognize new ones. |
Keywords
* Artificial intelligence * Fine tuning * Generalization * Overfitting * Pretraining
