| Popis: |
Chili is a vital crop in India, where it is widely grown for usage in food, medicine, and cosmetics. India is the world's largest producer and exporter of chilies, a key spice crop. For the leaf disease classification, Conventional Neural Networks (CNN) are used. EfficientNetB4 is a neural network architecture that has been demonstrated to achieve high accuracy on image classification tasks while using a small number of parameters. This model is calibrated for the function of detecting leaf disease. This research demonstrates the efficacy of using deep learning architectures for the detection of leaf diseases and highlights EfficientNetB4′s potential as a powerful tool for this task. We fine-tuned the EfficientNetB4 as EfficientLeafNetB4 and proved the most efficient in the detection of chili leaf disease detection compared to ResNet-50, DenseNet-121, MobileNet-V2, and VGG-16. Our proposed fine-tuned model produced better results as compared to the above-listed techniques.The proposed approach is simple to integrate into existing agricultural systems and can assist farmers in making timely and accurate crop health decisions. Furthermore, our work provides a roadmap for future research in this field, including the investigation of novel data augmentation techniques and regularization strategies and the exploration of other deep learning architectures.Sample abstract: The agriculture sector is essential to a country's Gross domestic product (GDP). Plants are essential because they provide people with sustenance. Most farmers in poor nations do manual farming. Plant diseases that are not identified in time can cause financial losses for farmers, damaging state and national economies on a huge scale. The present study investigates High-precision Multiclass Classification of Chili leaf Disease through Customized E ffecientNetB4 from Chili Leaf Images. This research represents the popular Indian crop leaf diseases like Up curl, Down curl, Gemini virus, Cercosporin leaf spot, healthy leaf, and their identification accuracies. We fine-tuned the EfficientNetB4 as EfficientLeafNetB4 and proved the most efficiency in the detection of the chili leaf disease detection compared to ResNet-50, DenseNet-121, MobileNet-V2 and VGG-16. Our proposed fine-tuned model produced better results as compared to the above listed techniques. |
