Cross-Level Multi-Instance Distillation for Self-Supervised Fine-Grained Visual Categorization

arXiv:2401.08860v2 Announce Type: replace
Abstract: High-quality annotation of fine-grained visual categories demands great expert knowledge, which is taxing and time consuming. Alternatively, learning fine-grained visual representation from enormous unlabeled images (e.g., species, brands) by self-supervised learning becomes a feasible solution. However, recent researches find that existing self-supervised learning methods are less qualified to represent fine-grained categories. The bottleneck lies in that the pre-text representation is built from every patch-wise embedding, while fine-grained categories are only determined by several key patches …

abstract annotation arxiv brands cs.cv distillation expert fine-grained images instance knowledge quality representation self-supervised learning solution supervised learning type visual