| 69 | 0 | 34 |
| 下载次数 | 被引频次 | 阅读次数 |
为了解决小样本学习中的类间差异性与类内一致性问题，提出一种结合自监督学习与原型网络的新方法SePNet(Self-supervised ProtoNet)。该方法利用自监督损失增强类间可分性，同时通过原型学习提升类内一致性。PlantVillage数据集上，在典型的5-way 1-shot和5-way 5-shot设置下对SePNet进行了评估，并与现有方法进行比较。实验结果表明，SePNet在小样本植物病害识别任务中具有较强的表现，验证了其有效性和应用潜力。
Abstract：To address the issues of inter-class discrepancy and intra-class consistency in few-shot learning, this paper proposes a novel approach called Self-supervised ProtoNet(SePNet), which integrates self-supervised learning with prototypical networks. The proposed method employs a self-supervised loss to enhance inter-class separability and utilizes prototype learning to improve intra-class consistency. We evaluate SePNet on the PlantVillage dataset under standard 5-way 1-shot and 5-way 5-shot settings, and compare its performance with existing methods. Experimental results demonstrate that SePNet exhibits strong performance in few-shot plant disease recognition, validating its effectiveness and potential for practical application.
[1] Atila ü,U?ar M,Akyol K,et al.Plant leaf disease classification using EfficientNet deep learning model[J].Ecological Informatics,2021,61:101182.
[2] Li Z,Liu F,Yang W,et al.A survey of convolutional neural networks:analysis,applications,and prospects[J].IEEE Transactions on Neural Networks and Learning Systems,2021:1-21.
[3] 王昕，刘爽，周长才.基于深度学习和磁共振图像的膝骨关节炎分类[J].长春工业大学学报，2023,44(1):45-51.
[4] Misra I,Maaten L.Self-supervised learning of pretext-invariant representations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020:6707-6717.
[5] Argüeso D,Picon A,Irusta U,et al.Few-shot learning approach for plant disease classification using images taken in the field[J].Computers and Electronics in Agriculture,2020,175:105542.
[6] Snell J,Swersky K,Zemel R.Prototypical networks for few-shot learning[C]//Advances in Neural Information Processing Systems,2017:30-38.
[7] Li L,Han J,Yao X,et al.DLA-matchNet for few-shot remote sensing image scene classification[J].IEEE Transactions on Geoscience and Remote Sensing,2020,59(9):7844-7853.
[8] Jeong T,Kim H.OOD-MAML:Meta-learning for few-shot out-of-distribution detection and classification[J].Advances in Neural Information Processing Systems,2020,33:3907-3916.
[9] Srivastava A,Wang T Y,Zhang P,et al.Memmap:Compact and generalizable meta-lstm models for memory access prediction[C]//Advances in Knowledge Discovery and Data Mining:24th Pacific-Asia Conference,PAKDD 2020,Singapore,May 11-14,2020,Proceedings,Part II 24.Springer International Publishing,2020:57-68.
[10] Li Z,Zhou F,Chen F,et al.Meta-sgd:Learning to learn quickly for few-shot learning[J].arXiv preprint arXiv:1707.09835,2017.
[11] Li X,Shi D,Diao X,et al.SCL-MLNet:Boosting few-shot remote sensing scene classification via self-supervised contrastive learning[J].IEEE Transactions on Geoscience and Remote Sensing,2021,60:1-12.
基本信息:
DOI：10.15923/j.cnki.cn22-1382/t.2025.6.07
中图分类号:TP18;S432
引用信息:
[1]高延生,刘冰,王梦辉.基于ProtoNet和自监督的植物叶子病害小样本分类[J].长春工业大学学报,2025,46(06):530-534.DOI:10.15923/j.cnki.cn22-1382/t.2025.6.07.
基金信息:
吉林省教育厅资助项目(JJKH20230765KJ,JJKH20230767KJ)
2025-12-03
2025-12-03