Test-time Adaptation with Slot-Centric Models
This addresses the challenge of parsing complex, unseen scenes into entities for applications like robotics and autonomous systems, representing an incremental advance by combining existing test-time adaptation and slot-centric methods.
The paper tackled the problem of scene decomposition for out-of-distribution scenes, where current visual detectors often fail, by proposing Slot-TTA, a semi-supervised slot-centric model adapted per test scene using gradient descent on reconstruction or cross-view synthesis objectives, resulting in substantial performance improvements over state-of-the-art supervised detectors and alternative adaptation methods.
Current visual detectors, though impressive within their training distribution, often fail to parse out-of-distribution scenes into their constituent entities. Recent test-time adaptation methods use auxiliary self-supervised losses to adapt the network parameters to each test example independently and have shown promising results towards generalization outside the training distribution for the task of image classification. In our work, we find evidence that these losses are insufficient for the task of scene decomposition, without also considering architectural inductive biases. Recent slot-centric generative models attempt to decompose scenes into entities in a self-supervised manner by reconstructing pixels. Drawing upon these two lines of work, we propose Slot-TTA, a semi-supervised slot-centric scene decomposition model that at test time is adapted per scene through gradient descent on reconstruction or cross-view synthesis objectives. We evaluate Slot-TTA across multiple input modalities, images or 3D point clouds, and show substantial out-of-distribution performance improvements against state-of-the-art supervised feed-forward detectors, and alternative test-time adaptation methods.