Advanced Disease Management: An Encoder-Decoder Approach for Tomato Black Mold Detection

The global transmission of tomato black mold disease is a critical issue due to its substantial impact on agricultural productivity and the overall quality of crops. The prompt highlights the critical role of prompt and precise illness identification in facilitating effective disease control and promoting ecologically friendly farming practices. This paper presents a novel methodology for the timely identification of black mold infestation in tomatoes. The proposed strategy utilizes a dense dilated convolution technique included inside an encoder-decoder architecture. The process of feature extraction utilizes the widely recognized VGG-16 model. Our model, on the other hand, has been trained and assessed using a rigorously curated dataset consisting of 25,000 tomato photos, which were collected by our team. The model has exceptional performance in accurately segmenting and precisely localizing contaminated areas, as indicated by the significantly high values of the Dice Similarity Coefficient (DSC) and Intersection over Union (IoU). Furthermore, the model's capacity to estimate ground truth boundaries is emphasized by the minimal Hausdorff Distance. Our research not only makes a significant contribution to the assurance of food safety but also advances the field of precision agriculture by providing farmers with a dependable method for disease identification. Our suggested method utilizes advanced computer vision techniques and deep learning (DL) models to enable the early and accurate detection of tomato black mold disease. This advancement has the potential to support sustainable crop management practices and provide significant benefits to the agricultural community.