Deep Batch-normalized eLU AlexNet For Plant Diseases Classification
Hmidi Alaeddine, Malek Jihene
Abstract
In early work, the automatic recognition problem of plant diseases relied on traditional machine learning techniques such as Multilayer Perceptrons (MLP) and Support Vector Machines (SVM). However, in recent years new approaches have moved towards the application of Deep Learning (DL) and convolutional neural network which is described as a dominant tool in this field. In this work, we introduce a model with an architecture based on the AlexNet model for the plant diseases classification from leaf images. We present a deeper version of AlexNet with size (3x3) convolution, normalization, regularization, and linear exponential unit (eLU) layers. The training and testing of the proposed model was performed on a PlantVillage dataset. This proposed model obtained precision and a high gain in convergence learning speed. It achieved 99.48% classification accuracy with 17.54x fewer parameters compared to AlexNet.