Apparatus and system for leaf disease classification and method of preparing thereof

Abstract

The present invention relates to an apparatus and a system for classification of leaf diseases and a method of preparing thereof, which belongs to the technical field of computer classification systems, specifically to machine learning algorithms applied to plant pathology. Said apparatus comprises a platform, and arm supported by said platform, a light acquisition device mounted onto said arm, characterized in that said apparatus further comprises a hub for data processing mounted on said arm. Said system comprises a single-board computer, a light-acquisition device electronically connected to it, a database comprising a leaf disease dataset comprising labeled data based on identified data and captured data stored in it, and a program installed in it, characterized in that said program further comprises a feature-extracting module comprising a convolutional neural network (CNN) trained with said database and a classifier module which is a multi-class classifier which computes inner product appended to said feature extracting module. Said method of preparing an apparatus and system for leaf disease classification comprises the steps of collecting raw data and preprocessing said raw data, wherein said collecting data comprises the steps of identifying leaf symptoms, capturing said leaf symptoms, and labeling said captured data, and wherein said pre-processing comprises the steps of resizing identified data, augmenting said resized data by rotation to several degrees, segmenting, and sharpening.