This work does not introduce a new method. Instead, we present an interesting finding that
questions the necessity of the inductive bias--locality in modern computer vision …

The existing contrastive learning methods mainly focus on single-grained representation
learning, eg, part-level, object-level or scene-level ones, thus inevitably neglecting the …

This paper represents a neat yet effective framework, named SemanticMIM, to integrate the
advantages of masked image modeling (MIM) and contrastive learning (CL) for general …

Contrastive learning has remarkable transfer learning capabilities. But many current
methods are pretrained based on instance-level or pixel-level pretext tasks, resulting in …