Machine Vision Based Carrot Disease Recognition

Abstract

Most of the people in Bangladesh lives on agriculture producing rice, jute, vegetables and so many crops. Fruits and vegetables have a large part of the agricultural sector which is still produced and supervised manually. Sometimes the diseases are not seen in the initial stage, which can severely affect our GDP. To ensure the freshness of fruits and vegetables modern image processing tools can help a lot. Experts can detect the defected fruits and vegetables by watching them with their eyes but the process is too long and not suitable for all the stores, farms, supermarkets or the exporters all around. There comes the blessings of new computer vision technologies with image processing techniques that can do a lot of works in a second. In this paper an automated approach is developed to detect defects of fruits and vegetables and recognize diseases by using machine vision based image processing techniques which is implemented in MATLAB including a machine learning algorithm with the Multiclass SVM classifier. There are many algorithms that can detect defects of fruits and vegetables hence, we separated the defected parts of the carrots using K-means clustering and then classified it with multiclass support vector machine classifier. Here, a supervised machine learning concept is implemented to recognize various carrot diseases. As the domain of this research model, carrot diseases are classified and 96% of accuracy is achieved which can certainly help in our agricultural science along with proper maintenance