We are happy and proud to announce that our paper titled “Perceptual Hashing Using Pretrained Vision Transformers” won the Best Paper Award at the IEEE Gaming, Entertainment and Media (GEM) 2024 conference! This was awarded by Technical Program Committee Chairs, based on the reviewers’ assesment.
The work presented in the paper was done in collaboration with the National Institute of Criminalistics and Criminology, and was done in the context of the master’s thesis of Jelle De Geest.
The paper explores how to use the power of large pretrained models to perform perceptual hashing, for image identification and deduplication. In simpler words, can we use this AI to automatically analyze a hard drive full of images - to aid forensic investigators? For example, to find illegal images such as child sexual abuse material. Or to group duplicate image variations (cropped, resized, compressed, …), such that investigators do not have to go through the same data multiple times. More information can be found here.
The paper was presented at IEEE GEM 2024, on 5 June 2024, in Turino, Italy, by Hannes Mareen.
Paper download: Perceptual Hashing Using Pretrained Vision Transformers
