Summary of Sfmvit: Slowfast Meet Vit in Chaotic World, by Jiaying Lin et al.
SFMViT: SlowFast Meet ViT in Chaotic World
by Jiaying Lin, Jiajun Wen, Mengyuan Liu, Jinfu Liu, Baiqiao Yin, Yue Li
First submitted to arxiv on: 25 Apr 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 proposed SFMViT network, a dual-stream spatiotemporal feature extraction network, improves model performance for spatiotemporal action localization in chaotic scenes. By leveraging ViT’s global feature extraction capabilities and SlowFast’s sequence modeling capabilities, the backbone of SFMViT enhances video feature extraction. Additionally, an anchor pruning strategy using confidence maximum heap helps filter out ineffective anchors, achieving a mAP of 26.62% on the Chaotic World dataset. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new network for spatiotemporal action localization in chaotic scenes. It uses a combination of ViT and SlowFast to extract features from videos. The model also includes an anchor pruning strategy to remove unwanted anchors. This approach leads to better performance than previous models, with a mAP of 26.62% on the Chaotic World dataset. |
Keywords
» Artificial intelligence » Feature extraction » Pruning » Spatiotemporal » Vit
