Abstract:As the global food security landscape continues to evolve, the need for accurate and reliable crop disease diagnosis has never been more pressing. To address global food security concerns, we extend the widely used PlantVillage dataset with additional apple leaf disease classes, enhancing diversity and complexity. Experimental evaluations on both original and extended datasets reveal that existing models struggle with the new additions, highlighting the need for more robust and generalizable computer vision models. Test F1 scores of 99.63% and 97.87% were obtained on the original and extended datasets, respectively. Our study provides a more challenging and diverse benchmark, paving the way for the development of accurate and reliable models for identifying apple leaf diseases under varying imaging conditions. The expanded dataset is available at <a class="link-external link-https" href="https://www.kaggle.com/datasets/akinyemijoseph/apple-leaf-disease-dataset-6-classes-v2" rel="external noopener nofollow">this https URL</a> enabling future research to build upon our findings.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to improve the accuracy and reliability of apple leaf disease classification in response to global food security issues. Specifically: 1. **Expand the existing dataset**: The paper expands the widely - used PlantVillage dataset by adding additional apple leaf disease categories (such as Alternaria leaf spot and powdery mildew), thus enhancing the diversity and complexity of the dataset. This helps to train more general - purpose and robust computer vision models. 2. **Improve model performance**: By introducing new disease categories, the research found that the existing models perform poorly when dealing with these newly added categories, highlighting the need to develop more powerful and general - purpose deep - learning models. The experimental results show that the F1 score on the original dataset is 99.63%, while the F1 score on the extended dataset is 97.87%. 3. **Provide a more challenging benchmark**: The extended dataset (PV - ALE) provides a more challenging and diverse benchmark, promoting the development of more accurate and reliable apple leaf disease identification models, especially under different imaging conditions. 4. **Promote future research**: By opening the extended dataset, researchers can conduct further research on this basis, promoting the progress in the field of apple leaf disease detection. ### Core contributions of the paper - **Comprehensive apple leaf image dataset**: High - quality annotated images with multiple apple leaf disease labels. - **Efficient CNN architecture**: A convolutional neural network architecture specifically designed for multi - class classification tasks. - **Strict evaluation metrics**: Evaluate the performance of the model in different scenarios, including cases of class imbalance. - **Superior accuracy**: Demonstrates higher accuracy than existing methods, indicating its potential in practical applications. ### Problems solved The paper mainly solves the following problems: - **Limitations of existing datasets**: Most existing datasets are small in size, have class imbalance and lack diversity. - **Insufficient model generalization ability**: Existing models perform poorly when facing new categories or complex backgrounds. - **Lack of publicly available large - scale datasets**: This hinders the training and verification of models and limits the progress of research. Through the solution of these problems, the paper provides new tools and methods for the field of apple leaf disease detection and promotes the progress of related technologies.

The PV-ALE Dataset: Enhancing Apple Leaf Disease Classification Through Transfer Learning with Convolutional Neural Networks

A Data Augmentation Method Based on Generative Adversarial Networks for Grape Leaf Disease Identification

Data Augmentation through Background Removal for Apple Leaf Disease Classification Using the MobileNetV2 Model

An Efficient Transfer Learning-based Approach for Apple Leaf Disease Classification

The Plant Pathology 2020 challenge dataset to classify foliar disease of apples

AFD-Net: Apple Foliar Disease multi classification using deep learning on plant pathology dataset

A Hybrid Model for Leaf Diseases Classification Based on the Modified Deep Transfer Learning and Ensemble Approach for Agricultural AIoT-Based Monitoring

Apple leaf disease identification via improved CycleGAN and convolutional neural network

Advancements in deep learning for accurate classification of grape leaves and diagnosis of grape diseases

Transfer learning for versatile plant disease recognition with limited data

A robust and light-weight transfer learning-based architecture for accurate detection of leaf diseases across multiple plants using less amount of images

Improving Wheat Leaf Disease Classification: Evaluating Augmentation Strategies and CNN-Based Models With Limited Dataset

Precision agriculture with YOLO-Leaf: advanced methods for detecting apple leaf diseases

Classification of Pear Leaf Diseases Based on Ensemble Convolutional Neural Networks

Improving plant disease classification using realistic data augmentation

Fruit and vegetable leaf disease recognition based on a novel custom convolutional neural network and shallow classifier

Real-Time Detection of Apple Leaf Diseases Using Deep Learning Approach Based on Improved Convolutional Neural Networks

Optimized classification model for plant diseases using generative adversarial networks

WITHDRAWN: Recognition of Apple Leaf Diseases using Deep Learning and Variances-Controlled Features Reduction

An ensemble of deep learning architectures for accurate plant disease classification

Identification of Multiple Diseases in Apple Leaf Based on Optimized Lightweight Convolutional Neural Network