Abstract: Global agriculture is greatly impacted by plant diseases, since different infections cause 20% to 40% of agricultural production to be lost each year. These losses are caused by bacteria, viruses, fungi, and other microorganisms that affect both economic stability and food security. For example, in the United States alone, the fungal infection Fusarium oxysporum causes losses of up to $1 billion yearly. Furthermore, academics and farmers alike are also concerned about the development of diseases like citrus greening and wheat rust. To mitigate these consequences and ensure long-lasting crop production, integrated approaches to disease control, like resistance breeding, cultural practices, and the prudent use of fungicide are crucial. Convolutional Neural Networks (CNNs), particularly those pre-trained on large datasets like ImageNet, have revolutionized identification of plant diseases by providing excellent accuracy and effectiveness in diagnosing various plant illnesses from images. This approach, which includes optimizing already trained models on plant disease datasets, reduces the requirement for big annotated datasets and computational resources, making it highly applicable for agricultural use. Google's Inception v3 model, known for its efficient architecture and use of inception modules, is widely used in plant disease diagnosis. It can precisely identify plant diseases through transfer learning by being pre-trained on ImageNet and fine-tuned on specific plant disease datasets. The Inception-ResNet v2 model, combining Inception architecture with residual networks, also excels in identification of plant diseases. Its deep structure captures detailed features from plant images, enabling accurate disease diagnosis. Like Inception v3, it uses transfer learning to generalize across various plant species and disease types, aiding in precision agriculture by facilitating early illness detection and timely intervention. This project aims to deploy and Compare the results of three models—Xception, Inception v3, and Inception-ResNet v2—in detecting fungal diseases in fruit plant
Keywords: Plant disease detection, CNN, Xception model, Inception V3 model, Inception ResNet V2 model
|
DOI:
10.17148/IJARCCE.2024.13616
