Summary of Freea: Human-object Interaction Detection Using Free Annotation Labels, by Yuxiao Wang et al.
FreeA: Human-object Interaction Detection using Free Annotation Labels
by Yuxiao Wang, Zhenao Wei, Xinyu Jiang, Yu Lei, Weiying Xue, Jinxiu Liu, Qi Liu
First submitted to arxiv on: 4 Mar 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 FreeA is a novel self-adaptation language-driven human-object interaction (HOI) detection method that leverages the adaptability of CLIP to generate latent HOI labels without requiring comprehensive annotated image datasets. The approach matches image features with HOI text templates and utilizes a priori knowledge-based mask method to suppress improbable interactions. Additionally, FreeA employs an interaction correlation matching method to refine generated HOI labels. Experimental results on two benchmark datasets demonstrate that FreeA achieves state-of-the-art performance among weakly supervised HOI models, outperforming existing methods by +8.58 mean Average Precision (mAP) on HICO-DET and +1.23 mAP on V-COCO. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary FreeA is a new way to detect when people interact with objects without needing lots of labeled pictures. It uses a special AI model called CLIP to figure out what’s happening in the picture. The method looks for patterns between people, objects, and actions, and then tries to match them up. This helps it learn what kinds of interactions are most likely to happen. FreeA is better than other methods at finding these interactions, making it a useful tool for studying how people use objects. |
Keywords
» Artificial intelligence » Mask » Mean average precision » Supervised
