An intelligent model of tomato leaf disease detection using adaptive hybrid segmentation and Residual ShuffleNet with attention mechanism

Tomato is considered an important agricultural crop that is widely used across various countries. Detecting tomato leaf disease has become an interesting research topic in the modern agricultural sector. Therefore, the research study investigates a new Tomato Leaf Disease Detection (TLDD) model based on several heuristic approaches. Moreover, the different phases of the study include the collection of tomato leaf images, segmenting abnormalities, optimal pixel selection, and tomato leaf disease detection. At first, the standard data sources are used to gather the required tomato leaf images. Furthermore, the gathered images are then undergone into a segmentation phase for segmenting the abnormalities using the Adaptive Hybrid Segmentation with Pixel Selection Method (AHSS-PSM) technique, in which the Region Growing (RG) and Watershed Algorithms (WA) are utilized. From the segmented images, the optimal pixels are selected with the aid of Adaptive Random Variable-based Osprey Optimization Algorithm (ARV-OOA). The optimally selected pixels are then provided as input into the Residual ShuffleNet with Attention Mechanism (ReShNet-AM) model to strengthen the tomato leaf disease detection performance. Therefore, an extensive experimental analysis is conducted to prove better performance of the suggested TLDD model against baseline detection models.