David Osowiechi (left) and Gustavo Vargas (right) are PhD students at ETS Montréal. Their supervisors and co-authors are Christian Desrosiers and Ismail Ben Ayed. Their work on test-time adaptation has been picked as a highlight paper, and David and Gustavo speak to us before their poster session this afternoon. NC-TTT: A Noise Contrastive Approach for Test-Time Training 8 DAILY CVPR Wednesday Highlight Presentation While deep learning models have demonstrated outstanding performance in vision tasks, they often struggle when confronted with domain shifts at test time – where the test data differs significantly from the training data. Test-Time Training (TTT), a specific
RkJQdWJsaXNoZXIy NTc3NzU=