Learning Based Photo Management on Smartphones

Nowadays, three major issues stand out in managing personal photo galleries: (1) the memory usage of many photos, (2) the clutter of photos in the gallery and challenges in gallery management, and (3) the difficulty in choosing the best photo from a series of consecutive shots. These issues not only negatively impact the user experience but also reduce device performance and make memory usage inefficient. In this paper, a method for grouping similar photos in a gallery using deep learning and unsupervised machine learning approaches is proposed. Then, the highest quality photo is selected from among similar photos using reference-free image quality metrics. The proposed approach will allow users to manage their digital content more effectively and use their storage snace more efficiently,