Summary of Crossfi: a Cross Domain Wi-fi Sensing Framework Based on Siamese Network, by Zijian Zhao et al.
CrossFi: A Cross Domain Wi-Fi Sensing Framework Based on Siamese Network
by Zijian Zhao, Tingwei Chen, Zhijie Cai, Xiaoyang Li, Hang Li, Qimei Chen, Guangxu Zhu
First submitted to arxiv on: 20 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Signal Processing (eess.SP)
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 CrossFi approach is a siamese network-based method that excels in both in-domain and cross-domain scenarios, including few-shot, zero-shot, and even new-class scenarios. To address the limitations of existing Wi-Fi sensing datasets, which often lead to over-fitting and poor performance in diverse scenarios, CrossFi employs a sample-similarity calculation network called CSi-Net that captures similarity information using an attention mechanism. This is complemented by a Weight-Net that generates a template for each class, enabling CrossFi to adapt to different scenarios. The approach achieves state-of-the-art performance across various scenarios, including gesture recognition with 98.17% accuracy in the in-domain scenario and 84.75% in one-shot new-class scenario. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary CrossFi is a new way to recognize gestures using Wi-Fi signals. This method can work well even when it’s not trained on the exact same data as what it’s seeing now. It does this by comparing the similarities between samples and generating templates for each class of gesture. The results show that CrossFi performs better than other methods in recognizing gestures, with high accuracy rates. |
Keywords
» Artificial intelligence » Attention » Few shot » Gesture recognition » One shot » Siamese network » Zero shot
