1. Hebei Key Laboratory of Measurement Technology and Instrument, Yanshan University, Qinhuangdao,Hebei 066004, China
2. School of Electrical Engineering, Yanshan University, Qinhuangdao, Hebei 066004, China
Abstract：In view of the fact that the existing single fine-grained identification model cannot identify the flower subclasses without training samples, a method of mapping fine-grained features to high-dimensional space automatic classification is proposed in combination with destruction and construction learning (DCL) and KNN to realize the classification of the subclasses without training samples. At the same time, in view of the similar characteristics of the same flower subclass and the possible imbalance of samples between classes, the loss function of the DCL model was improved. contrastive loss was used to increase the class spacing of the subclass, and focal loss balanced class Loss. Finally, the experiment was carried out on the peony flower with unbalanced 308 samples. The experimental results showed that the accuracy of the training sample subclass after the improved algorithm was 0.932, and the F1 score was 0.925, which was greatly improved compared with the original DCL algorithm. Meanwhile, the accuracy of the subclass without training sample was 0.903, and the F1 score was 0.888.
［1］Nilsback M E, Zisserman A. Automated flower classifica-tion over a large number of classes［C］//2008 Sixth Indian Conference on Computer Vision, Graphics & Ima-ge Processing. IEEE, Washington, USA, 2008: 722-729.
［2］Zhang C, Liu J, Liang C, et al. Image classification using Harr-like transformation of local features with cod-ing residuals ［J］. Signal Processing, 2013, 93 (8): 2111-2118.
［3］曹晓杰, 么娆, 严雨灵. 应用迁移学习的卷积神经网络花卉图像识别 ［J］. 计算机应用与软件, 2020, 37 (8): 142-148.
Cao X J, Mo Y, Yan Y L. The transfer learning convolutional neural network is applied for flower image recognition ［J］. Computer applications and software, 2020, 37 (8):142-148.
［4］秦敏. 基于深度学习的花卉图像分类识别模型研究 ［D］. 桂林:广西师范大学, 2020.
［5］Peng Y, He X, Zhao J. Object-part attention model for fine-grained image classification ［J］. IEEE Transactions on Image Processing, 2017, 27 (3): 1487-1500.
［6］Huang S, Xu Z, Tao D, et al. Part-stacked cnn for fine-grained visual categorization［C］//Proceedings of the IEEE conference on computer vision and pattern recogn-ition. Las Vegas, USA, 2016: 1173-1182.
［7］石龙杰, 周扬, 岑岗, 等. 基于OCT成像的淡水无核珍珠内部缺陷自动检测方法 ［J］. 计量学报, 2020, 41 (10): 1226-1233.
Shi L J, Zhou Y, Cen G, et al. Automatic Detection of Internal Defects in Freshwater Nuclefree Pearls Based on OCT ［J］. Acta Metrologica Sinica, 2020, 41 (10): 1226-1233.
［8］Cui Y, Song Y, Sun C, et al. Large scale fine-grained categorization and domain-specific transfer learning ［C］//Proceedings of the IEEE conference on computer vision and pattern recognition. Salt Lake City, USA, 2018: 4109-4118.
［9］Fu J, Zheng H, Mei T. Look closer to see better: Recurrent attention convolutional neural network for fine-grained image recognition［C］//Proceedings of the IEEE conference on computer vision and pattern recognition. Hawaii, USA, 2017: 4438-4446.
［10］Zhao B, Wu X, Feng J, et al. Diversified visual attention networks for fine-grained object classification ［J］. IEEE Transactions on Multimedia, 2017, 19 (6): 1245-1256.
［11］王金甲, 周雅倩, 郝智. 基于注意力模型的多传感器人类活动识别 ［J］. 计量学报, 2019, 40 (6): 958-969.
Wang J J, Zhou Y Q, Hao Z, Multi-sensor human activity recognition based on attention model ［J］. Acta Metrologica Sinica, 2019, 40 (6): 958-969.
［12］Lin T-Y, Goyal P, Girshick R, et al. Focal loss for dense object detection ［C］//Proceedings of the IEEE international conference on computer vision. Venice, Italy, 2017: 2980-2988.
［13］Wu Z, Xiong Y, Yu S X, et al. Unsupervised feature learning via non-parametric instance discrimination ［C］//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA, 2018: 3733-3742.
［14］Hadsell R, Chopra S, LeCun Y. Dimensionality reduction by learning an invariant mapping ［C］//2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR06). IEEE, New York, USA, 2006: 1735-1742.
［15］Dong X, Shen J. Triplet loss in siamese network for object tracking ［C］//Proceedings of the European Con-ference on Computer Vision (ECCV). Munich, Germany, 2018: 459-474.
［16］Chen Y, Bai Y, Zhang W, et al. Destruction and construction learning for fine-grained image recognition ［C］//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Long Beach, CA, USA, 2019: 5157-5166.
［17］Keller J M, Gray M R, Givens JA, et al. A fuzzy k-nearest neighbor algorithm［J］,1985 (4): 580-585.
［18］Wah C, Branson S, Welinder P, et al. The caltech-ucsd birds-200-2011 dataset［DB］, https://xueshu.baidu.com/usercenter/paper/show?paperid=7b01fb907ef9c2dc0fb4c1e70307b3e4,2011.
［19］张玉燕, 李永保, 温银堂, 等. 基于CT图像的金属点阵结构内部缺陷检测方法 ［J］. 计量学报, 2020, 41 (5): 544-550.
Zhang Y Y, Li Y B, Wen Y T, et al. Method for Detecting Interal Defects of Metal Lattice Structure Based on CT Image ［J］. Acta Metrologica Sinica, 2020,41 (5): 544-550.
［20］樊凤杰, 轩凤来, 白洋, 等. 基于LLE-RF的中药三维荧光光谱分类识别 ［J］. 计量学报, 2020, 41 (2): 263-268.
Fan F J, Xuan F L, Bai Y, et al. Classification Recognition of Three Dimensional Fluorescence Spec-trum Traditioal Chinese Medicine Based on LLE-RF ［J］. Acta Metrologica Sinica, 2020,41 (2): 263-268.
［21］Lin T Y, Roychowdhury A, Maji S. Bilinear cnn models for fine-grained visual recognition ［C］// Proceedings of the IEEE international conference on computer vision. Santiago, Chile, 2015: 1449-1457.
2021, Vol. 42 
