Semi-supervised few-shot learning approach for plant diseases recognition

• Published in: Plant methods Vol. 17; no. 1; p. 68
• Main Authors: Li, Yang, Chao, Xuewei
• Format: Journal Article
• Language: English
• Published: London BioMed Central 27.06.2021; BioMed Central Ltd; BMC
• Subjects: Accelerating image-based plant phenotyping and pattern recognition: deep learning or few-shot learning?; Agricultural production; Agriculture; Biological Techniques; Biomedical and Life Sciences; Classification; confidence interval; Confidence intervals; data collection; Datasets; Deep learning; Diagnosis; domain; Domains; Labeling; Learning; Leaves; Life Sciences; Machine learning; Plant diseases; Plant Sciences; Plants; Recognition; Self-adaption; Semi-supervised learning; Supervised learning; Transfer learning; Unmanned aerial vehicles; China; Deep learning; Self-adaption; Transfer learning; Classification
• ISSN: 1746-4811; 1746-4811
• DOI: 10.1186/s13007-021-00770-1

Background Learning from a few samples to automatically recognize the plant leaf diseases is an attractive and promising study to protect the agricultural yield and quality. The existing few-shot classification studies in agriculture are mainly based on supervised learning schemes, ignoring unlabeled data's helpful information. Methods In this paper, we proposed a semi-supervised few-shot learning approach to solve the plant leaf diseases recognition. Specifically, the public PlantVillage dataset is used and split into the source domain and target domain. Extensive comparison experiments considering the domain split and few-shot parameters (N-way, k-shot) were carried out to validate the correctness and generalization of proposed semi-supervised few-shot methods. In terms of selecting pseudo-labeled samples in the semi-supervised process, we adopted the confidence interval to determine the number of unlabeled samples for pseudo-labelling adaptively. Results The average improvement by the single semi-supervised method is 2.8%, and that by the iterative semi-supervised method is 4.6%. Conclusions The proposed methods can outperform other related works with fewer labeled training data.