A Method of Plant Disease Detection Analysis From Image Object Extraction Based on the Mask R-CNN Model

Plant leaf infections are a common threat to global production in the long and short term affecting not only many farmers but also consumers worldwide. Early detection and treatment of plant leaf diseases are essential to promote healthy growth of agricultural plants and ensure adequate supply and health security for the world's geometric growth (population). Detection of plant leaf diseases using computer-aided technologies is widespread nowadays. In this work, a method for improving the quality of plant images in order to boost the performance of leaf-based disease detection algorithms is proposed. To this end, an efficient Deep Learning (D L) modified mask region convolutional neural network(R-CNN mask) is proposed for autonomous segmentation and detection of the object to be analyzed in this research. The proposed work exploited the features learned by the Mask R-CNN model at various processing hierarchies. We used images of corn leaves and apples in fields from the standard Plant Village repository for object localization. We have through an improvement of the generated code singular images of the detected objects and an overall accuracy of 98.89 % on the validation sets was achieved. which allowed us to obtain very interesting results.