PlantEIRNet: Plant Leaf Disease Detection Deep Neural Networks Using Farmland Images

The productivity and quality of crops are greatly influenced by various diseases and pests. These are one of a kind of natural disasters which affect the natural growth of plants and can even cause the death of the plants ultimately affecting the yield of the crops. Detection of these diseases is henceforth really important. Such plant disease detected via image processing or object detection techniques. Deep Learning techniques are highly used for these kinds of tasks since they significantly outperform the conventional approaches. This research paper proposed an enhanced Inception ResNet-V2 for detecting plant diseases and is called as PlantEIRNet. Also, various state-of-the-art convolutional neural networks such as InceptionV3, ResNet-50, VGG-16, AlexNet, MobileNet-V3 are compared with the proposed method. Plant diseases are detected inspite of real-world settings such as illumination, occlusion and varied viewpoints. Three datasets namely, PlantLD-11 dataset, PlantVillage dataset and FusedPD dataset are assessed for their performance using PlantEIRNet. The proposed enhanced Inception ResNet-V2 outperformed compared to other Convolutional Neural Networks such as Inception V3, ResNet50, VGG-16 and Inception ResNet-V2.