Orthogonal features improve unsupervised object grouping accuracy
OrthoRF: Exploring Orthogonality in Object-Centric Representations
Computer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
Summary
Identifying and grouping objects in images without labels is challenging. The authors propose a new method that makes certain parts of the model's internal features orthogonal, leading to clearer distinctions between objects. This approach helps the model handle overlapping objects, noise, and occlusion better than previous methods. It also removes the need for extra clustering steps, making object discovery more straightforward and interpretable.
What this means in practice
- •For computer vision engineers: Improve automatic object detection systems to identify and separate overlapping and occluded objects without labeled data.
- •For robotics developers: Enhance robots’ scene understanding by enabling more reliable object grouping in noisy or challenging visual environments.
Authors
Despoina Touska, Bastiaan Onne Fagginger Auer, Alexandru Onose, Tejaswi Kasarla, Luis Armando Pérez Rey, Maximilian Lipp, Lyubov Amitonova, Martin R. Oswald, Pascal Cerfontaine
Abstract
Neural synchrony is hypothesized to help the brain organize visual scenes into structured multi-object representations. In machine learning, synchrony-based models analogously learn object-centric representations by storing binding in the phase of complex-valued features. Rotating Features (RF) instantiate this idea with vector-valued activations, encoding object presence in magnitudes and affiliation in orientations. We propose Orthogonal Rotating Features (OrthoRF), which enforces orthogonality in RF's orientation space via an inner-product loss and architectural modifications. This yields sharper phase alignment and more reliable grouping. In evaluations of unsupervised object discovery, including settings with overlapping objects, noise, and out-of-distribution tests, OrthoRF matches or outperforms current models while producing more interpretable representations, and it eliminates the post-hoc clustering required by many synchrony-based approaches. Unlike current models, OrthoRF also recovers occluded object parts, indicating stronger grouping under occlusion. Overall, orthogonality emerges as a simple, effective inductive bias for synchrony-based object-centric learning.