ArticlePublisher preview available

Identification of apple leaf diseases using C-Grabcut algorithm and improved transfer learning base on low shot learning

Springer Nature
Multimedia Tools and Applications
Authors:
Suyun Lian
Suyun Lian
Suyun Lian
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Lixin Guan
Lixin Guan
Lixin Guan
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Jihong Pei
Jihong Pei
Jihong Pei
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Gui Zeng
Gui Zeng
Gui Zeng
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Show all 5 authorsHide
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

Plant disease control is an indispensable research topic in the field of agriculture. Different apple leaf diseases may have similar manifestations, and it is time-consuming and laborious to rely on manual means. In this paper, we propose an apple leaf disease classification algorithm for a small number of samples, which is based on the C-Grabcut image segmentation algorithm proposed in this paper and the improved EfficientNetB4 transfer learning algorithm. Firstly, data augmentation is used to expand the samples, which effectively solves the problems of insufficient samples and unbalanced sample categories. Then the leaves are extracted from the images using the C-Grabcut algorithm to reduce the interference brought by the background. Finally, the improved Vgg16, ResNet50, EfficientNetB0, EfficientNetB4 and EfficientNetB7 transfer learning algorithms are used to classify leaves into four categories: rust, scab, multiply and healthy. The experimental results show that the improved EfficientB4 algorithm works best with an average accuracy of 98% and the Kappa value of 0.98. In addition, the C-Grabcut algorithm reduces the training time from 153 to 73 s during an epoch, allowing the proposed algorithms to be deployed on devices with lower computing power and memory.
Figures - available from: Multimedia Tools and Applications
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Springer Nature.
Learn more
Content available from Multimedia Tools and Applications 
This content is subject to copyright. Terms and conditions apply.
Vol.:(0123456789)
Multimedia Tools and Applications (2024) 83:27411–27433
https://doi.org/10.1007/s11042-023-16602-4
1 3
Identification ofapple leaf diseases using C‑Grabcut
algorithm andimproved transfer learning base onlow shot
learning
SuyunLian1,2 · LixinGuan1 · JihongPei2 · GuiZeng1 · MengshanLi1
Received: 29 March 2022 / Revised: 12 August 2023 / Accepted: 21 August 2023 /
Published online: 1 September 2023
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2023
Abstract
Plant disease control is an indispensable research topic in the field of agriculture. Different
apple leaf diseases may have similar manifestations, and it is time-consuming and labori-
ous to rely on manual means. In this paper, we propose an apple leaf disease classification
algorithm for a small number of samples, which is based on the C-Grabcut image segmen-
tation algorithm proposed in this paper and the improved EfficientNetB4 transfer learn-
ing algorithm. Firstly, data augmentation is used to expand the samples, which effectively
solves the problems of insufficient samples and unbalanced sample categories. Then the
leaves are extracted from the images using the C-Grabcut algorithm to reduce the interfer-
ence brought by the background. Finally, the improved Vgg16, ResNet50, EfficientNetB0,
EfficientNetB4 and EfficientNetB7 transfer learning algorithms are used to classify leaves
into four categories: rust, scab, multiply and healthy. The experimental results show that
the improved EfficientB4 algorithm works best with an average accuracy of 98% and the
Kappa value of 0.98. In addition, the C-Grabcut algorithm reduces the training time from
153 to 73s during an epoch, allowing the proposed algorithms to be deployed on devices
with lower computing power and memory.
Keywords Apple leaf diseases· C-Grabcut· Transfer learning· Segmentation
1 Introduction
The world population is increasing and and the demand for food is increasing [1]. There are
three main reasons affecting crop production: failure to effectively prevent plant diseases
[2], loss of arable land [3], and climate change [4]. Plant disease control is an indispensable
research topic in the field of agriculture. By the implementation of control measures, crop
yields has been increased significantly [5, 6]. Some diseases were caused by environmental
* Lixin Guan
lxguan@gnnu.edu.cn
1 School ofPhysics andElectronic Information, Gannan Normal University, Ganzhou341000,
China
2 Electronics Engineering Department With CIE, Shenzhen University, Shenzhen518000, China
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
... This reduction enables the network to fulfill the deployment needs of platforms with limited storage and computational power. Lian et al. [18] proposed an improved GrabCut algorithm, called C-GrabCut, to remove backgrounds. Five pre-trained models, namely Vgg16, ResNet50, EfficientNetB0, EfficientNetB4, and EfficientNetB7, were used to classify apple leaf images from Plant Pathology dataset into four classes: rust, scab, multiply, and healthy. ...
... In the second strategy, the number of images in each class is increased to match the size of the largest class. This can performed by generating synthetic images through transformations like rotation, flipping, and scaling [18]. In our study, we did not consider the underrepresented class of multiple diseases, and adopted the first strategy by randomly selecting 516 images for classes "scab" and "rust". ...
Data Augmentation through Background Removal for Apple Leaf Disease Classification Using the MobileNetV2 Model
Preprint
Full-text available
  • Nov 2024
The advances in computer vision made possible by deep learning technology are increasingly being used in precision agriculture to automate the detection and classification of plant diseases. Symptoms of plant diseases are often seen on their leaves. The leaf images in existing datasets have been collected either under controlled conditions or in the field. The majority of previous studies have focused on identifying leaf diseases using images captured in controlled laboratory settings, often achieving high performance. However, methods aimed at detecting and classifying leaf diseases in field images have generally exhibited lower performance. The objective of this study is to evaluate the impact of a data augmentation approach that involves removing complex backgrounds from leaf images on the classification performance of apple leaf diseases in images captured under real world conditions. To achieve this objective, the lightweight pre-trained MobileNetV2 deep learning model was fine-tuned and subsequently used to evaluate the impact of expanding the training dataset with background-removed images on classification performance. Experimental results show that this augmentation strategy enhances classification accuracy. Specifically, using the Adam optimizer, the proposed method achieved a classification accuracy of 98.71% on the Plant Pathology database, representing an approximately 3% improvement and outperforming state-of-the-art methods. This demonstrates the effectiveness of background removal as a data augmentation technique for improving the robustness of disease classification models in real-world conditions.
View
... Trained on a dataset of high-resolution orange leaf images from Mexico, the DenseNet model achieved an accuracy of 99.50% in classifying 12 distinct anomaly types, underscoring its potential for precise and scalable citrus disease management. Lian et al. [14] introduces an Inception-ResNet-V2 model for the highly accurate identification (99.91%) of 38 leaf diseases across 11 crop species utilizing the PlantVillage dataset consisting of 54,305 coloured leaf images (healthy and diseased). The model surpasses other deep learning architectures in performance and integrates a local interpretable framework to highlight disease-specific image regions, thereby enhancing transparency and fostering user trust. ...
A hybrid multi-optimizer approach using CNN and GB for accurate prediction of citrus fruit diseases
Article
Full-text available
  • Feb 2025
Efficient prediction of citrus fruit diseases is essential for maintaining orchard health and productivity. Traditional diagnostic methods, often relying on manual inspection, are labor-intensive and prone to inaccuracies. Deep learning techniques, especially Convolutional Neural Networks (CNNs), offer an automated and accurate alternative. This study introduces a novel model integrating CNN with Gradient Boosting (GB) and optimized using the Nesterov-Accelerated Adaptive Moment Estimation (Nadam) optimizer to enhance prediction accuracy. The model employs a custom CNN architecture combined with GB, leveraging Nadam for faster convergence and improved performance. Trained on a dataset of 3,000 citrus fruit images sourced from Kaggle, the model follows a structured process of preprocessing, feature extraction, integration of GB with CNN, and optimal prediction. Comparative analysis using metrics such as accuracy, precision, F1 score, and recall demonstrates the model's effectiveness, achieving an accuracy of 98.03% and precision of 98.04%. This robust approach addresses limitations of traditional methods by enabling automated feature extraction and reliable disease prediction. The proposed CNN-GB-Nadam model significantly enhances efficiency and reliability, providing a valuable tool for protecting citrus fruit health and improving orchard management practices.
View
Lightweight Plant Disease Classification Combining GrabCut Algorithm, New Coordinate Attention, and Channel Pruning
Article
Full-text available
  • Jun 2022
  • NEURAL PROCESS LETT
In this paper, a lightweight convolutional neural network is proposed for the classification of plant diseases, containing 63 classes of states for 11 plant species. The different context of experimental data and data in the real environment, insufficient accuracy of the model classification, and oversized model are three main problems of deep learning techniques applied to agricultural production. In this paper, we mainly focus on these three problems. First, the GrabCut algorithm is adopted to unify the background of the experimental data and the real data to black, allowing the trained model to have the same good effect when applied in practice. Then, we propose a new coordinate attention block to improve the classification accuracy of convolutional neural networks and empirically demonstrate the effectiveness of our approach with several state-of-the-art CNN models. Finally, channel pruning is applied to the trained model, which reduces the model size and computational effort by 85.19%%\% and 92.15%%\% respectively with little change in the model accuracy, making it better suited for agricultural platforms with lower memory and computational capacity.
View
Identification of Apple Tree Leaf Diseases Based on Deep Learning Models
Article
Full-text available
  • Jun 2020
Early diagnosis and accurate identification of apple tree leaf diseases (ATLDs) can control the spread of infection, to reduce the use of chemical fertilizers and pesticides, improve the yield and quality of apple, and maintain the healthy development of apple cultivars. In order to improve the detection accuracy and efficiency, an early diagnosis method for ATLDs based on deep convolutional neural network (DCNN) is proposed. We first collect the images of apple tree leaves with and without diseases from both laboratories and cultivation fields, and establish dataset containing five common ATLDs and healthy leaves. The DCNN model proposed in this paper for ATLDs recognition combines DenseNet and Xception, using global average pooling instead of fully connected layers. We extract features by the proposed convolutional neural network then use a support vector machine to classify the apple leaf diseases. Including the proposed DCNN, several DCNNs are trained for ATLDs recognition. The proposed network achieves an overall accuracy of 98.82% in identifying the ATLDs, which is higher than Inception-v3, MobileNet, VGG-16, DenseNet-201, Xception, VGG-INCEP. Moreover, the proposed model has the fastest convergence rate, and a relatively small number of parameters and high robustness compared with the mentioned models. This research indicates that the proposed deep learning model provides a better solution for ATLDs control. It could be also integrated into smart apple cultivation systems.
View
Rice plant disease classification using color features: a machine learning paradigm
Article
Full-text available
  • Oct 2020
In traditional practices, detection of rice plant diseases by experts is a subjective matter whereas by testing in the laboratory is time-consuming. As a consequence, it causes reduction on agricultural production and economic loss to farmers. To overcome this, there is a demand to develop fast and effective systems to detect and classify rice plant diseases. Therefore, the development of image-based automated systems for classification of rice plant diseases is an interesting growing research area in the agriculture domain. Color is one of the important features to classify rice plant diseases. In this study, we have presented an image-based rice plant disease classification approach using color features only. We have explored 14 different color spaces and extracted four features from each color channel leading to 172 features. Moreover, the performance of seven different classifiers have been compared and demonstrated that a highest classification accuracy of 94.65% has been achieved using support vector machine (SVM) classifier. Training and testing of models were performed on the dataset that consists of 619 images. This dataset was collected from the real agriculture field that belongs to four classes: (a) Bacterial Leaf Blight (BLB), (b) Rice Blast (RB), (c) Sheath Blight (SB) and (d) Healthy Leave (HL). The encouraging results of this paper show that color features can play an important role in developing rice plant disease classification system and enable the farmers to take preventive measures resulting in better product quality and quantity.
View
Apple Leaf Diseases Recognition Based on An Improved Convolutional Neural Network
Article
Full-text available
Scab, frogeye spot, and cedar rust are three common types of apple leaf diseases, and the rapid diagnosis and accurate identification of them play an important role in the development of apple production. In this work, an improved model based on VGG16 is proposed to identify apple leaf diseases, in which the global average poling layer is used to replace the fully connected layer to reduce the parameters and a batch normalization layer is added to improve the convergence speed. A transfer learning strategy is used to avoid a long training time. The experimental results show that the overall accuracy of apple leaf classification based on the proposed model can reach 99.01%. Compared with the classical VGG16, the model parameters are reduced by 89%, the recognition accuracy is improved by 6.3%, and the training time is reduced to 0.56% of that of the original model. Therefore, the deep convolutional neural network model proposed in this work provides a better solution for the identification of apple leaf diseases with higher accuracy and a faster convergence speed.
View
Meta-Seg: A Generalized Meta-Learning Framework for Multi-Class Few-Shot Semantic Segmentation
Article
Full-text available
  • Nov 2019
Semantic segmentation performs pixel-wise classification for given images, which can be widely used in autonomous driving, robotics, medical diagnostics and etc. The recent advanced approaches have witnessed rapid progress in semantic segmentation. However, these supervised learning based methods rely heavily on large-scale datasets to acquire strong generalizing ability, such that they are coupled with some constraints. Firstly, human annotation of pixel-level segmentation masks is laborious and time-consuming, which causes relatively expensive training data and make it hard to deal with urgent tasks in dynamic environment. Secondly, the outstanding performance of the above data-hungry methods will decrease with few available training examples. In order to overcome the limitations of the supervised learning semantic segmentation methods, this paper proposes a generalized meta-learning framework, named Meta-Seg. It consists of a meta-learner and a base-learner. Specifically, the meta-learner learns a good initialization and a parameter update strategy from a distribution of few-shot semantic segmentation tasks. The base-learner can be any semantic segmentation models theoretically and can implement fast adaptation (that is updating parameters with few iterations) under the guidance of the meta-learner. In this work, the successful semantic segmentation model FCN8s is integrated into Meta-Seg. Experiments on the famous few-shot semantic segmentation dataset PASCAL5i prove Meta-Seg is a promising framework for few-shot semantic segmentation. Besides, this method can provide with reference for the relevant researches of meta-learning semantic segmentation.
View
View
World Population Prospects 2019: Highlights
Technical Report
  • Jun 2019
Available at: https://population.un.org/wpp/Publications/Files/WPP2019_Highlights.pdf The 2019 revision of the World Population Prospects is the twenty-sixth edition of the United Nations population estimates and projections. It presents population estimates from 1950 to the present for 235 countries or areas, underpinned by analyses of historical demographic trends. This latest assessment considers the results of 1,690 national population censuses conducted between 1950 and 2018, as well as information from vital registration systems and from 2,700 nationally representative sample surveys. The 2019 revision also presents population projections to the year 2100 that reflect a range of plausible outcomes at the global, regional and country levels.
View
A Mutually Supervised Graph Attention Network for Few-Shot Segmentation: The Perspective of Fully Utilizing Limited Samples
Article
  • Mar 2022
Fully supervised semantic segmentation has performed well in many computer vision tasks. However, it is time-consuming because training a model requires a large number of pixel-level annotated samples. Few-shot segmentation has recently become a popular approach to addressing this problem, as it requires only a handful of annotated samples to generalize to new categories. However, the full utilization of limited samples remains an open problem. Thus, in this article, a mutually supervised few-shot segmentation network is proposed. First, the feature maps from intermediate convolution layers are fused to enrich the capacity of feature representation. Second, the support image and query image are combined into a bipartite graph, and the graph attention network is adopted to avoid losing spatial information and increase the number of pixels in the support image to guide the query image segmentation. Third, the attention map of the query image is used as prior information to enhance the support image segmentation, which forms a mutually supervised regime. Finally, the attention maps of the intermediate layers are fused and sent into the graph reasoning layer to infer the pixel categories. Experiments are conducted on the PASCAL VOC- 5i5^i dataset and FSS-1000 dataset, and the results demonstrate the effectiveness and superior performance of our method compared with other baseline methods.
View
View
Identification of cash crop diseases using automatic image segmentation algorithm and deep learning with expanded dataset
Article
  • Oct 2020
  • COMPUT ELECTRON AGR
Using deep learning methods to identify cash crop diseases has become a current hotspot in the field of plant disease identification. However, recent studies have demonstrated that the complex background information of crop images from practical application and insufficient training data can cause the wrong recognition of deep learning. To address this problem, in this paper we present an identification method of cash crop diseases using automatic image segmentation and deep learning with expanded dataset. An Automatic Image Segmentation Algorithm(AISA) based on the GrabCut algorithm is designed to remove the background information of images automatically while retaining the disease spots. It doesn‘t need to select the object manually during image processing and is of much lower time cost compared with the GrabCut algorithm. The MobileNet Convolutional Neural Network(CNN) model is selected as the deep learning model and plenty of crop images from the Internet and practical planting bases are added to expand the public dataset PlantVillage for the purpose of improving the generalization ability of MobileNet. The images are processed by the AISA before they can be used for extracting disease features, which reduces calculations significantly and ensures that the disease features of the crop leaf can be extracted accurately. Moreover, we design a cash crop disease identification system for mobile smart devices. The experimental results show that the system has a correct recognition rate of more than 80% for the 27 diseases of 6 crops described in this paper and then has a high value of practical application.
View