Automatic Leaf Recognition Based on Deep Semi-Supervised Learning

被引：0

作者：

Wu H. ^{[1
]}

Xiao F. ^{[1
]}

Shi Z. ^{[1
]}

Wen Z. ^{[1
]}

机构：

[1] College of Computer Science and Software Engineering, Shenzhen University, Shenzhen

来源：

Jisuanji Fuzhu Sheji Yu Tuxingxue Xuebao/Journal of Computer-Aided Design and Computer Graphics | 2023年 / 35卷 / 10期

关键词：

convolutional neural network; deep learning; feature extraction; plant leaf recognition; semi-supervised learning;

D O I：

10.3724/SP.J.1089.2023.19487

中图分类号：

学科分类号：

摘要：

Automatic leaf recognition algorithms have been widely used in the fields of plant teaching and ecological protection. However, due to the wide variety of plants and the small differences between classes, traditional deep learning methods require a large amount of data annotations to obtain better training results. Therefore, a semi-supervised deep learning method for plant leaf automatic identification is proposed. Firstly, based on the idea of consistent regularization, both of explicit and implicit data perturbation processes are designed to improve the quality of data perturbation. Secondly, by utilizing a deep feature extraction network DenseNet, the ability of fine detail recognition can be effectively improved. Finally, based on the simulated annealing training strategy, the abnormal data in the training process is filtered to alleviate the over-fitting phenomenon. Extensive experiments are conducted on a public dataset MalayaKew-D3 with 5284 leaf images and a private dataset LeafSZU-2021 with 18000 leaf images. Compared with the fully supervised learning methods, the proposed method can still achieve 92.36% to 96.85% accuracy of plant leaf recognition by using only 30% to 50% labeled data. Compared with the latest semi-supervised spherical mean clustering model under the same amount of data annotations, the average recognition accuracy of the proposed method improves by 2.95%, where the number of model parameters also decreases by 38.12%, and the recognition speed improves by 61.51%. © 2023 Institute of Computing Technology. All rights reserved.

引用

页码：1469 / 1478

页数：9

共 26 条

[1]

Mora C, Tittensor D P, Adl S, Et al., How many species are there on Earth and in the ocean?, PLoS Biology, 9, 8, (2011)

[2]

Laine S, Aila T., Temporal ensembling for semi-supervised learning

[3]

Wu H S, Xiang Y K, Liu J J, Et al., Automatic leaf recognition based on deep convolutional networks, Proceedings of International Conference on Neural Information Processing, pp. 505-515, (2017)

[4]

Lee S H, Chan C S, Mayo S J, Et al., How deep learning extracts and learns leaf features for plant classification, Pattern Recognition, 71, pp. 1-13, (2017)

[5]

Sun Y, Liu Y, Wang G, Et al., Deep learning for plant identification in natural environment, Computational Intelligence and Neuroscience, 2017, (2017)

[6]

Hu J, Chen Z B, Yang M, Et al., A multiscale fusion convolutional neural network for plant leaf recognition, IEEE Signal Processing Letters, 25, 6, pp. 853-857, (2018)

[7]

Zhang S W, Lei Y K, Zhang C L, Et al., Semi-supervised orthogonal discriminant projection for plant leaf classification, Pattern Analysis and Applications, 19, 4, pp. 953-961, (2016)

[8]

Wang Z B, Sun X G, Yang Z K, Et al., Leaf recognition based on DPCNN and BOW, Neural Processing Letters, 47, 1, pp. 99-115, (2018)

[9]

Alamoudi S, Hong X, Wei H., Plant leaf recognition using texture features and semi-supervised spherical K-means clustering, Proceedings of 2020 International Joint Conference on Neural Networks, pp. 1-8, (2020)

[10]

Holub A, Welling M, Perona P., Exploiting unlabelled data for hybrid object classification, Proceedings of Neural Information Processing Systems, Workshop Inter-Class Transfer, (2005)

← 1 2 3 →