Adaptive deep clustering integrating DINOv2 embeddings, graph attention, and bio-inspired optimization

Mai Abdrabo, Hossam Refaat, Mohammed Abdallah, Osama Farouk · 2025

Read Adaptive deep clustering integrating DINOv2 embeddings, graph attention, and bio-inspired optimization by Mai Abdrabo, Hossam Refaat, Mohammed Abdallah, Osama Farouk (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12749389.txt.

Description

Grouping photos without labels is hard when groups blur into each other. Abdrabo and co-authors present a 2025 method that links three tools into one loop. A vision model first pulls rich features from each image. A graph network then sharpens links between like images.

A bat-inspired search step picks how many groups to use and tunes the setup as it goes. Each step feeds back into the others, rather than running alone. The authors add two new scores that judge how clean and tight the groups are. Tests in the paper compare the loop with older ways of grouping images. The aim is groups that stay clear even on messy photo sets.

Catalog record

CREATORMai Abdrabo, Hossam Refaat, Mohammed Abdallah, Osama Farouk
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE119 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC12749389

Browse the full library · About the Tower of Babel