Raspberry Pi based crop and weed classification using Machine Learning Algorithm

Abstract

Farming with fun has uplifted the agritourism in our country. The technological development and support have made farming a passion nowadays. Farming is not only important to cater the needs of the growing population but also to develop agritourism. The major challenge in crop production like irrigation, de-weeding, etc can be handled with good improvement in quality crops through the implementation of automated systems in monitoring and controlling application. With more Techies joining into the farming wishing to lead a stress-free life, automation. The work presented aims at identifying the weeds from the desired crop using the learning algorithm. The weed location identified can be used as the inputs to the de-weeding Robots. Besides, the work can equally be applicable in intercropping process, irrigation management, quality assurance in crops, etc. The input images of the crop to be classified are taken using the camera interfaced with Raspberry pi. Classification is done using the supervised learning algorithm with a set of target classes. The model is trained to recognize the target class. The accuracy of the classification varies from 75% to 100% as the proportion of the training and test data set varies from 60-40% to 80-20%.