On Equivariant and Invariant Learning of Object Landmark Representations
This work addresses the challenge of learning geometric equivariant and invariant representations for object landmark discovery, which is important for computer vision tasks like matching and few-shot regression, but it appears incremental as it builds on existing contrastive learning approaches.
The paper tackles the problem of unsupervised discovery of object landmarks in images by combining instance-discriminative and spatially-discriminative contrastive learning, resulting in a method that surpasses prior state-of-the-art on standard benchmarks and a new challenging one.
Given a collection of images, humans are able to discover landmarks by modeling the shared geometric structure across instances. This idea of geometric equivariance has been widely used for the unsupervised discovery of object landmark representations. In this paper, we develop a simple and effective approach by combining instance-discriminative and spatially-discriminative contrastive learning. We show that when a deep network is trained to be invariant to geometric and photometric transformations, representations emerge from its intermediate layers that are highly predictive of object landmarks. Stacking these across layers in a "hypercolumn" and projecting them using spatially-contrastive learning further improves their performance on matching and few-shot landmark regression tasks. We also present a unified view of existing equivariant and invariant representation learning approaches through the lens of contrastive learning, shedding light on the nature of invariances learned. Experiments on standard benchmarks for landmark learning, as well as a new challenging one we propose, show that the proposed approach surpasses prior state-of-the-art.