Summary of Rethinking Data Augmentation For Robust Lidar Semantic Segmentation in Adverse Weather, by Junsung Park et al.
Rethinking Data Augmentation for Robust LiDAR Semantic Segmentation in Adverse Weather
by Junsung Park, Kyungmin Kim, Hyunjung Shim
First submitted to arxiv on: 2 Jul 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 paper addresses the issue of LiDAR semantic segmentation performance decline in adverse weather conditions, such as fog or rain. Existing methods have simulated adverse weather or used universal data augmentation during training, but lacked a detailed analysis of how weather affects performance. The authors identify two key factors: geometric perturbation due to refraction and point drop due to energy absorption and occlusions. They propose new strategic data augmentation techniques, including Selective Jittering (SJ) and Learnable Point Drop (LPD), which mimic the effects of adverse weather conditions. These methods enhance robustness against weather conditions, achieving a notable 39.5 mIoU on the SemanticKITTI-to-SemanticSTF benchmark, improving the baseline by 8.1%p and establishing a new state-of-the-art. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a problem with LiDAR technology, which helps self-driving cars see their surroundings. When it’s foggy or rainy outside, LiDAR gets worse at recognizing objects. The authors figured out what makes this happen: some light gets bent and others get blocked. They created new ways to make the system better by adding fake “weather” to the training data. This helps the system learn to recognize things even when it’s bad weather. Their method works really well, beating previous best results. |
Keywords
» Artificial intelligence » Data augmentation » Semantic segmentation
