A study of different disease detection and classification techniques using deep learning for cannabis plant

Abstract

In this paper, different models for disease detection and classification are studied for cannabis plants. Cannabis plants are used for medical and recreational purposes with its recent legalization in some places. Cannabis farmers face problems in cultivation of the crop since it’s susceptible to multiple disorders. With early detection of the disease in the crop it is possible to prevent large waste of yield in the crop. A real dataset is considered for disease detection and classification purposes which is a combination of text and image data and that has been collected over a period of one and a half years (Feb 2018-August 2019). The models used in this study are Fast Region Convolutional Neural Network(F-RCNN), MobileNet Single Shot Multibox Detector(MobileNet-SSD), You Only Look Once(YOLO) and Residual Network-50 Layers (ResNet50). It is found that the MobileNet-SSD provided the best accuracy amongst all the object detection models that are studied and has a lesser training time as well. ResNet 50 is used for identifying the number of images required for a good fit without having to label first and then studied for the object detection models.