Automated Detection of Onion Health Using Deep Learning

The increasing global demand for sustainable agriculture has led to the adoption of artificial intelligence in farming practices. This paper presents an automated system for detecting the health of onions using deep learning techniques, specifically leveraging the MobileNetV2 architecture. The model was trained and validated on a curated dataset of healthy and unhealthy onion images, employing data augmentation to enhance generalizability. Our pipeline includes preprocessing, feature extraction, and classification, achieving a test accuracy of over 91.11% on unseen data. The implementation emphasizes efficiency, ensuring compatibility with low-resource environments. This work also provides a comparative evaluation of model performance metrics and highlights potential applications in precision agriculture. The proposed system demonstrates significant potential to assist farmers in early disease detection, reducing crop losses and enhancing productivity.