Summary of Impacts Of Color and Texture Distortions on Earth Observation Data in Deep Learning, by Martin Willbo et al.
Impacts of Color and Texture Distortions on Earth Observation Data in Deep Learning
by Martin Willbo, Aleksis Pirinen, John Martinsson, Edvin Listo Zec, Olof Mogren, Mikael Nilsson
First submitted to arxiv on: 7 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: 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 This research paper investigates how different visual characteristics of Earth observation (EO) data affect the predictions made by deep learning models trained for land cover classification and change detection. Specifically, it examines how convolutional and transformer-based U-net models respond to various color- and texture-based distortions in EO data during inference. The study reveals that these state-of-the-art models are generally more sensitive to texture than color distortions, which can have implications for model development and robustness within the EO domain. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research looks at how well deep learning models work when they’re shown different kinds of Earth observation data. It tests what happens when these models see data that’s been changed in certain ways, like being more colorful or having a different texture. The study found that these models are usually better at handling changes to the color of the data rather than its texture. This can help us make better models for using Earth observation data to classify land cover and detect changes. |
Keywords
* Artificial intelligence * Classification * Deep learning * Inference * Transformer
