Abstract
Smartphone photography has improved considerably over the years, particularly in low-light conditions. Thanks to better sensors, advanced noise reduction and night modes, smartphones can now capture detailed wide-field astronomical images, from the Moon to deep sky objects. These innovations have transformed them into pocket telescopes, making astrophotography more accessible than ever. In this paper, we present AstroSmartphoneDataset, a dataset of wide-field astronomical images collected since 2021 with various smartphones, and we show how we have used these images to train Deep Learning models for trails detection.