Multi-Step Preprocessing With UNet Segmentation and Transfer Learning Model for Pepper Bell Leaf Disease Detection

Agricultural production is a cornerstone of national economies, and the prevalence of plant diseases poses a significant threat to crop yields. Timely disease detection is essential to mitigate these risks. However, manual plant observation methods are labor-intensive and time-consuming, necessitating a shift toward automated solutions. This study addresses the pressing problem of plant disease identification by leveraging advanced image processing techniques. This research begins with a comprehensive analysis of the pepper bell leaf disease dataset. Through a series of meticulously designed image processing steps, the dataset is normalized, enhancing its quality and consistency. Building upon this preprocessing, the UNET segmentation technique in conjunction with the InceptionV3 transfer learning model is employed. This novel approach yields exceptional results, with 99.48% accuracy, 99.97% precision, 99.99% recall, and 99.98% F1 scores. To objectively assess the significance of the proposed model, the performance is benchmarked against existing state-of-the-art models. The findings demonstrate the superiority of the proposed approach in the domain of plant disease identification. By automating the detection process, this research not only enhances efficiency but also enables early disease detection, thereby potentially contributing to the agricultural sector to identify crop disease and manage it efficiently.