For any queries, please contact at srinadhml99@gmail.com
This work has been accepted to the Elsevier Journal of Astronomy and Computing, 2022 https://www.sciencedirect.com/science/article/abs/pii/S2213133722000488.
For more details on code and trained models, please visit our LFOVIA lab webpage here [https://github.com/lfovia/Attention-Augmented-Cosmic-Ray-Detection-in-Astronomical-Images](url).
We have developed an interactive web application to automatically identify and visualize large cosmic ray (CR) hits for close inspection. The app provides:
- Live CR mask visualization over the input FITS images
- Manual editing support for refining the CR masks
- Dynamic threshold tuning for adapting to various observational datasets
Figure: Interactive CR mask visualization interface of the Attention U-Net web application.
This tool is handy for astronomers to fine-tune detection thresholds and interactively inspect CR contamination patterns.
We are actively adding new features — the final version, featuring transformer-based CR segmentation and analysis, will be released soon!
Stay tuned for the next major update with improved model performance, multi-image batch support, and enhanced web interactivity.
If you find this project useful in your research, please consider citing:
@article{bhavanam2022cosmic,
title = {Cosmic Ray rejection with attention augmented deep learning},
author = {Bhavanam, Srinadh Reddy and Channappayya, Sumohana S and Srijith, P. K. and Desai, Shantanu},
journal = {Astronomy and Computing},
volume = {40},
pages = {100625},
year = {2022},
publisher = {Elsevier}
}