Analysis of Few-Shot Techniques for Fungal Plant Disease Classification and Evaluation of Clustering Capabilities Over Real Datasets

Abstract

Plant fungal diseases are one of the most important causes of crop yield losses. Therefore, plant disease identification algorithms have been seen as a useful tool to detect them at early stages to mitigate their effects. Although deep-learning based algorithms can achieve high detection accuracies, they require large and manually annotated image datasets that is not always accessible, specially for rare and new diseases. This study focuses on the development of a plant disease detection algorithm and strategy requiring few plant images (Few-shot learning algorithm). We extend previous work by using a novel challenging dataset containing more than 100,000 images. This dataset includes images of leaves, panicles and stems of five different crops (barley, corn, rape seed, rice, and wheat) for a total of 17 different diseases, where each disease is shown at different disease stages. In this study, we propose a deep metric learning based method to extract latent space representations from plant diseases with just few images by means of a Siamese network and triplet loss function. This enhances previous methods that require a support dataset containing a high number of annotated images to perform metric learning and few-shot classification. The proposed method was compared over a traditional network that was trained with the cross-entropy loss function. Exhaustive experiments have been performed for validating and measuring the benefits of metric learning techniques over classical methods. Results show that the features extracted by the metric learning based approach present better discriminative and clustering properties. Davis-Bouldin index and Silhouette score values have shown that triplet loss network improves the clustering properties with respect to the categorical-cross entropy loss. Overall, triplet loss approach improves the DB index value by 22.7% and Silhouette score value by 166.7% compared to the categorical cross-entropy loss model. Moreover, the F-score parameter obtained from the Siamese network with the triplet loss performs better than classical approaches when there are few images for training, obtaining a 6% improvement in the F-score mean value. Siamese networks with triplet loss have improved the ability to learn different plant diseases using few images of each class. These networks based on metric learning techniques improve clustering and classification results over traditional categorical cross-entropy loss networks for plant disease identification.

Meta‐learning shows great potential in plant disease recognition under few available samples

“…Few-Shot Learning Metric-based (Li and Chao, 2020;Arg¨ueso et al, 2020;Monowar et al, 2022;Egusquiza et al, 2022;Li and Yang, 2020;Zhang et al, 2022a;Cai et al, 2021) Model-based ) Optimization-based Tseng et al, 2022;Zhai et al, 2022;Saleem et al, 2020…”

“…(a) The applications and methods of meta‐learning: (1) meta‐learning is combined with target detection techniques and thus used in the diagnosis of plant diseases; (2) few‐shot recognition of plant diseases is the main application direction of meta‐learning; (3) meta‐learning can be applied to segmentation tasks with plant diseases; (4) based on the disease recognition, the use of meta‐learning to discriminate the degree of disease (Bock et al., 2021) is an important research direction; (5) combining meta‐learning with time series models and thus for the analysis of plant disease mechanisms (Kundu et al., 2019); and (6) the application of meta‐learning to the construction of plant disease knowledge graphs (Zhang, Wang, et al., 2022) is a research direction of interest. (b) The algorithm procedure of meta‐learning: (7) analysis of few‐shot techniques for fungal plant disease classification and evaluation of clustering capabilities over real datasets (Egusquiza et al., 2022); (8) active learning with point supervision for cost‐effective panicle detection in cereal crops (Chandra et al., 2020); (9) rectified meta‐learning from noisy labels for robust image‐based plant disease diagnosis (Zhai et al., 2022); and (10) few‐shot cotton leaf spots disease classification based on metric learning (Liang, 2021).…”

“…These questions result in a costly collection of real existing images with disease annotations and difficult disease identification tasks for real scenes. For example, Egusquiza et al (2022) present a few-shot learning method for disease diagnoses of cereal crops in a real environment, such as wheat, barley, and rice. As shown in Figure 2(b,7), this method can achieve the recognition accuracy of traditional deep-learning methods with only a small number of training samples, which solves the problem of accurate recognition in the case of scarce disease samples.…”

A Machine Learning Method to Identify Umami Peptide Sequences by Using Multiplicative LSTM Embedded Features

“…The loss function is defined simply as follows: where y is the sample label, which takes the value of 1 if the sample is a positive case and 0 otherwise, and is the probability that the model predicts that the sample is a positive case. In general, the lower the value of the cross-entropy loss function, the higher the classification effect [ 52 , 53 , 54 , 55 ].…”

Transfer learning for versatile plant disease recognition with limited data

“…Therefore, limited dataset , a situation where a few labeled images are accessible for some classes in the training process is one of the main issues in the literature ( Fan et al., 2022 ). To facilitate this issue, many algorithms and strategies are proposed, such as data augmentation ( Mohanty et al., 2016 ; Xu et al., 2022b ; Olaniyi et al., 2022 ), transfer learning ( Mohanty et al., 2016 ; Too et al., 2019 ; Chen J. et al., 2020 ; Xing and Lee, 2022 ; Zhao et al., 2022 ), few-shot learning ( Afifi et al., 2020 ; Egusquiza et al., 2022 ), and semi-supervised learning ( Li and Chao, 2021 ).…”