Summary of Iterative Geometry Encoding Volume For Stereo Matching, by Gangwei Xu et al.
Iterative Geometry Encoding Volume for Stereo Matching
by Gangwei Xu, Xianqi Wang, Xiaohuan Ding, Xin Yang
First submitted to arxiv on: 12 Mar 2023
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: None
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 paper proposes a new deep network architecture called Iterative Geometry Encoding Volume (IGEV-Stereo) for stereo matching, which builds upon the Recurrent All-Pairs Field Transforms (RAFT) framework. The IGEV-Stereo combines geometry and context information with local matching details to update the disparity map iteratively. To improve convergence speed, it uses Generalized Encoding Volume (GEV) to regress an accurate starting point for ConvGRUs iterations. The proposed method achieves state-of-the-art performance on KITTI 2015 and 2012 (Reflective), outperforming other published methods while being the fastest among top-ranked approaches. Additionally, IGEV-Stereo demonstrates strong cross-dataset generalization and high inference efficiency. The authors also extend their approach to multi-view stereo (MVS) with competitive accuracy on the DTU benchmark. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about a new way to match images that are slightly different because of how they were taken. The old method, RAFT, was good but not perfect. So, the researchers created a new model called IGEV-Stereo that combines information from the images with details about how they’re related. This helps the model do better and faster than before. They tested it on some big datasets and it came out on top. The model also works well when trying to match multiple images at once. |
Keywords
* Artificial intelligence * Generalization * Inference
