Adaptive snake optimization-enabled deep learning-based multi-classification using leaf images

Plants role is crucial for the preservation of earth's environment and ecology by maintaining healthy surroundings. Because of their proximity during the entire year, the leaves are crucial for identifying a plant. Plant’s diseases may disturb the leaf in between planting and the harvesting stage, resulting a significant loss in crop production and commercial benefit. As a result, leaf disease detection is difficult task in the field of agriculture. However, it necessitates a large workforce, additional processing time, and good knowledge of plant diseases. To address this problem, an optimized deep learning (DL) approach is employed for leaf type classification as well as disease detection. Initially, the leaf image is preprocessed with anisotropic filtering before being segmented using Mask-R-CNN. The multiclassification of leaf type and disease detection is carried out here. In leaf type classification process, the SqueezeNet with adaptive snake optimizer (ASO) is used. In disease detection process, the deep Qnet with proposed ASO is utilized. The proposed ASO-SqueezeNet technique performed better in classification process, with accuracy, precision, sensitivity, specificity and F-measure are 0.924, 0.888, 0.930, 0.935, and 0.909. Furthermore, the performance of ASO-deep Qnet in disease detection process is 0.919, 0.870, 0.922, 0.924, and 0.911, respectively.