Accurate Disease Detection in Apple Fruit Images using Transfer Learning: A Novel Approach for Agricultural Sustainability

Apple fruit disease poses a significant challenge to the agriculture industry due to its negative impact on orchard yield and economic growth. The current method of identifying diseased apples is a manual inspection process that is time-consuming and requires significant labor. To address this issue, the proposed work focuses on transfer learning based models namely ResNet, MobileNet, Inception, EfficientNet and Xception, to detect disease in apple images. The transfer learning based approaches aim to quickly identify infections and prevent further spread caused by environmental factors, allowing for efficient selection of healthy fruits while also replacing the manual inspection system to speed up the process and improve accuracy and efficiency. The transfer learning models are evaluated on apple images dataset. The obtained research results show that the Xception model outperforms the rest of the transfer learning models with significant margin.