Summary of Zero-shot Pupil Segmentation with Sam 2: a Case Study Of Over 14 Million Images, by Virmarie Maquiling et al.
Zero-Shot Pupil Segmentation with SAM 2: A Case Study of Over 14 Million Images
by Virmarie Maquiling, Sean Anthony Byrne, Diederick C. Niehorster, Marco Carminati, Enkelejda Kasneci
First submitted to arxiv on: 11 Oct 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
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 A medium-difficulty summary: This paper explores the potential of SAM 2, a vision foundation model, in advancing gaze estimation and eye tracking technologies. The authors demonstrate the model’s ability to significantly reduce annotation time, lower technical barriers through ease of deployment, and enhance segmentation accuracy. By utilizing its zero-shot segmentation capabilities with minimal user input, the researchers tested SAM 2 on over 14 million eye images from diverse datasets, including virtual reality setups and a unified dataset recorded using wearable eye trackers. The model achieves competitive mean Intersection over Union (mIoU) scores of up to 93% without fine-tuning in pupil segmentation tasks, matching the performance of domain-specific models trained solely on eye images. The authors also provide their code and segmentation masks for these widely used datasets to promote further research. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A low-difficulty summary: This paper looks at how a special kind of AI model called SAM 2 can help make gaze estimation and eye tracking technologies better. Gaze estimation is the ability to figure out where someone is looking, and eye tracking is the ability to track people’s eye movements. The authors found that SAM 2 can reduce the time it takes to prepare data for these tasks, making it easier for researchers to work on this area of study. They also tested SAM 2 on a huge amount of data from different sources, including virtual reality setups and a big unified dataset recorded using special glasses that track eye movements. The results show that SAM 2 can do just as well as more specialized models without needing extra training. |
Keywords
» Artificial intelligence » Fine tuning » Sam » Tracking » Zero shot
