Test-time Adaptation with Slot-Centric Models

TLDR: Current SOTA methods for scene understanding, though impressive, often fail to decompose out-of-distribution scenes. In our ICML paper, Slot-TTA (http://slot-tta.github.io) we find that optimizing per test sample over reconstruction loss improves scene decomposition accuracy. Problem Statement: In machine learning, we often assume the train and test split are IID samples from the same distribution. However, this doesn’t hold true in reality. In fact, there is a distribution shift happening all the time! For example on the left, we visualize …

computer vision deep learning machine learning research