Summary of Enhancing Autonomous Vehicle Perception in Adverse Weather Through Image Augmentation During Semantic Segmentation Training, by Ethan Kou et al.
Enhancing Autonomous Vehicle Perception in Adverse Weather through Image Augmentation during Semantic Segmentation Training
by Ethan Kou, Noah Curran
First submitted to arxiv on: 14 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 presents a study on improving semantic segmentation models for autonomous vehicle navigation and localization in various weather conditions. The authors focus on domain adaptation, where they train a segmentation model using clear-weather images and then apply image augmentation techniques to simulate different weather effects. They use the CARLA simulator to collect a dataset of 1200 images from 10 towns, and also create a set of 1200 images with random weather effects. The authors find that applying augmentations significantly improves segmentation performance under weathered night conditions, but there is still room for improvement in the domain adaptation approach. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper explores how to make autonomous vehicles better at understanding what’s around them, even in bad weather like fog or rain. Right now, most training data is just images taken on sunny days, which doesn’t help with adverse weather conditions. The authors think that if they can add fake weather effects (like random rain or fog) to the training data, it will help the model learn to recognize things more easily in different weather. They tested this idea using a simulator and found that it does improve performance, but there’s still work to be done to make it even better. |
Keywords
* Artificial intelligence * Domain adaptation * Semantic segmentation
