Cross-Level Multi-Instance Distillation for Self-Supervised Fine-Grained Visual Categorization
IEEE Transactions on Image Processing · 2025 · 34 · 2954-2969
DOI: 10.1109/tip.2025.3567834 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Image Processing, 2025. Verfasst von Qi Bi und 4 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 79 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
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 investigations find that existing self-supervised learning methods are less qualified to represent fine-grained categories. The bottleneck lies in that the pre-trained class-agnostic representation is built from every patch-wise embedding, while fine-grained categories are only determined by several key patches of an image. In this paper, we propose a Cross-level Multi-instance Distillation (CMD) framework to tackle this challenge. Our key idea is to consider the importance of each image patch in determining the fine-grained representation by multiple instance learning. To comprehensively learn the relation between informative patches and fine-grained semantics, the multi-instance knowledge distillation is implemented on both the region/image crop pairs from the teacher and student net, and the region-image crops inside the teacher / student net, which we term as intra-level multi-instance distillation and inter-level multi-instance distillation. Extensive experiments on several commonly used datasets, including CUB-200-2011, Stanford Cars and FGVC Aircraft, demonstrate that the proposed method outperforms the contemporary methods by up to 10.14% and existing state-of-the-art self-supervised learning approaches by up to 19.78% on both top-1 accuracy and Rank-1 retrieval metric. Source code is available at https://github.com/BiQiWHU/CMD.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Bi, Q., Ji, W., Yi, J., Zhan, H., & Xia, G. S. (2025). Cross-Level Multi-Instance Distillation for Self-Supervised Fine-Grained Visual Categorization. IEEE Transactions on Image Processing, 34, 2954-2969. https://doi.org/10.1109/tip.2025.3567834
Vancouver
Bi Q, Ji W, Yi J, Zhan H, Xia GS. Cross-Level Multi-Instance Distillation for Self-Supervised Fine-Grained Visual Categorization. IEEE Transactions on Image Processing. 2025;34:2954-2969. doi:10.1109/tip.2025.3567834
BibTeX
@article{Bi2025,
author = {Bi, Qi and Ji, Wei and Yi, Jingjun and Zhan, Haolan and Xia, Gui-Song},
title = {Cross-Level Multi-Instance Distillation for Self-Supervised Fine-Grained Visual Categorization},
journal = {IEEE Transactions on Image Processing},
year = {2025},
volume = {34},
pages = {2954--2969},
doi = {10.1109/tip.2025.3567834}
}
EUVIMED