Abstract:
The spectre of devastating citrus diseases looms large over global agriculture, threatening
yields and food security. Early and precise detection is crucial to mitigate crop losses and
ensure food security. This research explores the potential of deep learning for real-time
citrus disease detection, pitting four formidable convolutional neural network (CNN)
architectures against each other: CNN, DenseNet, ResNet, and Inception. Utilizing a
publicly available citrus disease dataset from Kaggle encompassing five classes, VGG16
emerges as the champion, achieving a remarkable 99% accuracy. This groundbreaking
performance paves the way for a VGG16-powered real-time disease detection system.
Such a system, patrolling citrus groves like a vigilant digital eye, could revolutionize
disease management. Early intervention translates to minimized losses, benefiting farmers
and the agricultural industry. Furthermore, by minimizing reliance on harmful pesticides,
VGG16-based detection offers a path to a healthier environment and safer food production.
This research paves the path for further exploration of deep learning applications in
precision agriculture, contributing to a more secure and sustainable food system, where
citrus thrives not just for economic prosperity, but for the well-being of our planet and the
generations to come.
