Summary of Edgenat: Transformer For Efficient Edge Detection, by Jinghuai Jie et al.
EdgeNAT: Transformer for Efficient Edge Detection
by Jinghuai Jie, Yan Guo, Guixing Wu, Junmin Wu, Baojian Hua
First submitted to arxiv on: 20 Aug 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 A transformer-based edge detector called EdgeNAT is proposed, leveraging the hierarchical structure of Dilated Neighborhood Attention Transformer (DiNAT) to efficiently capture both global and local features. This one-stage approach combines DiNAT as an encoder with a novel SCAF-MLA decoder that utilizes inter-spatial and inter-channel relationships. Extensive experiments on multiple datasets demonstrate state-of-the-art performance, including ODS F-measure and OIS F-measure of 86.0% and 87.6% on the BSDS500 dataset, surpassing the current state-of-the-art EDTER by 1.2%, 1.1%, 1.7%, and 1.6%, respectively. The approach also achieves a throughput of 20.87 FPS on an RTX 4090 GPU with single-scale input. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary EdgeNAT is a new way to find the edges in images using transformers, which are really good at learning features from data. This method uses another technique called DiNAT to look at both big and small details in the image. It then uses a special kind of math to make sure it gets the edges right. The results show that EdgeNAT is better than other methods at finding edges, especially on images with depth information. |
Keywords
* Artificial intelligence * Attention * Decoder * Encoder * Transformer
