I’ve been advising Gabriel Bertocco since he was an undergrad student and now as a Ph.D. Student. It’s such a pleasure to see how a great researcher he is becoming! Proud of you Gabs! Very proud! 😍❤️
Today, he had his first article (as the main author) published in the top notch IEEE Transactions on Information Forensics and Security.
Curious to see his work? It’s all about unsupervised and self-supervised learning! Take a look 😁
Unsupervised Domain Adaptation (UDA) is an important approach to Person Re-Identification (ReID). Our main contribution is a novel UDA-based ReID method that optimizes a simple loss function, with only one hyper-parameter, and takes advantage of the diversity of cameras to select reliable data for the unsupervised training at each iteration. This new strategy regularizes the model, avoiding overfitting to the target domain.