SatStreaks: Towards Supervised Learning for Delineating Satellite Streaks from Astronomical Images
by Susrita Chatterjee, Prachi Kudeshia, Nikolaus Kollo, Muhammad Altaf Agowun, Jiju Peethambaran, and Yasushi Akiyama
Published onMay 28, 2024
SatStreaks: Towards Supervised Learning for Delineating Satellite Streaks from Astronomical Images
·
ABSTRACT
Delineation of satellite streaks in astronomical images is an important aspect of ground based space studies. While deep learning algorithms show promise, training and validation of deep learning models for satellite streak segmentation is challenging due to the limited availability of large-scale, annotated datasets. We introduce SatStreaks, a dataset comprising of 3,130 densely annotated, real images of satellite streaks captured through ongoing citizen science projects. We utilize SatStreaks to develop a U-Net based model for the streak segmentation and conduct an experimental evaluation of data-driven image segmentation algorithms. The satellite streak segmentation codebase consisting of various deep learning models, and the SatStreak dataset has been made publicly available (https://github.com/jijup/SatStreaks) to facilitate the advancement of computer vision algorithms for space studies.
Creative Commons Attribution 4.0 International License (CC-BY 4.0)