Abstract：In ultra-fine-grained visual classification(Ultra-FGVC), the inter-class differences are extremely subtle while the intra-class variations remain considerable. To address these issues, an ultra-fine-grained visual classification network based on dynamic confusion discovery(DCD-Net) is proposed. DCD-Net is composed of two key modules: the dynamic confusion discovery module(DCDM) and the confusion-aware dual contrastive learning module(CDCLM). DCDM constructs a confusion affinity matrix by analyzing the predicted probability distributions of the model. The globally optimal class pairing relationships are derived to identify the most confusable class pairs at the current training stage. CDCLM focuses on these identified confusion pairs to optimize the feature space from two perspectives: preserving intra-class feature consistency and enlarging the feature margin between easily confused classes. A collaborative mechanism is formed by the two modules through a confusion pairing table, enabling the network to dynamically adjust its learning focus throughout the training and continuously concentrate on the most indistinguishable class boundaries. Experiments on five ultra-fine-grained datasets and one fine-grained dataset demonstrate that DCD-Net achieves high recognition accuracy and strong generalization ability.
[1] YU X H, WANG J, ZHAO Y, et al. Mix-ViT: Mixing Attentive Vision Transformer for Ultra-Fine-Grained Visual Categorization. Pattern Recognition, 2023, 135. DOI: 10.1016/j.patcog.2022.109131. [2] YU X H, WANG J, GAO Y S.CLE-ViT: Contrastive Learning Encoded Transformer for Ultra-Fine-Grained Visual Categorization // Proc of the 32nd International Joint Conference on Artificial Intelligence. San Francisco, USA: IJCAI, 2023: 4531-4539. [3] YU X H, ZHAO Y, GAO Y S.SPARE: Self-Supervised Part Erasing for Ultra-Fine-Grained Visual Categorization. Pattern Recognition, 2022, 128. DOI: 10.1016/j.patcog.2022.108691. [4] LIN T Y, ROYCHOWDHURY A, MAJI S.Bilinear CNN Models for Fine-Grained Visual Recognition // Proc of the IEEE International Conference on Computer Vision. Washington, USA: IEEE, 2015: 1449-1457. [5] HAN J W, YAO X W, CHENG G, et al. P-CNN: Part-Based Convolutional Neural Networks for Fine-Grained Visual Categorization. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(2): 579-590. [6] ZHENG H L, FU J L, MEI T, et al. Learning Multi-attention Con-volutional Neural Network for Fine-Grained Image Recognition // Proc of the IEEE International Conference on Computer Vision. Washington, USA: IEEE, 2017: 5219-5227. [7] HE J, CHEN J N, LIU S, et al. TransFG: A Transformer Architecture for Fine-Grained Recognition. Proceedings of the AAAI Confe-rence on Artificial Intelligence, 2022, 36(1): 852-860. [8] CHEN Y, BAI Y L, ZHANG W, et al. Destruction and Construction Learning for Fine-Grained Image Recognition // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2019: 5152-5161. [9] PU Y F, HAN Y Z, WANG Y L, et al. Fine-Grained Recognition with Learnable Semantic Data Augmentation. IEEE Transactions on Image Processing, 2024, 33: 3130-3144. [10] JIANG X, TANG H, GAO J Y, et al. Delving into Multi-modal Prompting for Fine-Grained Visual Classification. Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 38(3): 2570-2578. [11] YU X H, ZHAO Y, GAO Y S, et al. MaskCOV: A Random Mask Covariance Network for Ultra-Fine-Grained Visual Categorization. Pattern Recognition, 2021, 119. DOI: 10.1016/j.patcog.2021.108067. [12] CHEN Q P, JIAO L, WANG F M, et al. Integrating Foreground-Background Feature Distillation and Contrastive Feature Learning for Ultra-Fine-Grained Visual Classification. Pattern Recognition, 2024, 150. DOI: 10.1016/j.patcog.2024.110339. [13] CHEN T, KORNBLITH S, NOROUZI M, et al. A Simple Framework for Contrastive Learning of Visual Representations[C/OL].[2026-02-23]. https://arxiv.org/pdf/2002.05709v1. [14] CHEN X L, FAN H Q, GIRSHICK R, et al. Improved Base-lines with Momentum Contrastive Learning[C/OL].[2026-02-23]. https://arxiv.org/pdf/2003.04297. [15] GRILL J B, STRUB F, ALTCHÉ F, et al. Bootstrap Your Own Latent-A New Approach to Self-Supervised Learning // Proc of the 34th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2020: 21271-21284. [16] YUN S, HAN D, CHUN S, et al. CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features // Proc of the IEEE/CVF International Conference on Computer Vision. Wa-shington, USA: IEEE, 2019: 6022-6031. [17] CHOE J, SHIM H.Attention-Based Dropout Layer for Weakly Supervised Object Localization // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2019: 2214-2223. [18] DEVRIES T, TAYLOR G W.Improved Regularization of Convolutional Neural Networks with Cutout[C/OL].[2026-02-23]. https://arxiv.org/pdf/1708.04552. [19] TOUVRON H, CORD M, DOUZE M, et al. Training Data-Efficient Image Transformers & Distillation Through Attention // Proc of the 38th International Conference on Machine Learning. San Diego, USA: JMLR, 2021: 10347-10357. [20] SUN H B, HE X T, PENG Y X.SIM-Trans: Structure Information Modeling Transformer for Fine-Grained Visual Categorization // Proc of the ACM International Conference on Multimedia. New York, USA: ACM, 2022: 5853-5861. [21] DOSOVITSKIY A, BEYER L, KOLESNIKOV A, et al. An Image Is Worth 16×16 Words: Transformers for Image Recognition at Scale[C/OL].[2026-02-23]. https://arxiv.org/pdf/2010.11929. [22] LIU Z, LIN Y T, CAO Y, et al. Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows // Proc of the IEEE/CVF International Conference on Computer Vision. Washington, USA: IEEE, 2021: 9992-10002.
2026, Vol. 39 
