Deep learning-based rice leaf diseases detection using yolov5

Muhammad Juman Jhatial; Riaz Ahmed Shaikh; Noor Ahmed Shaikh; Samina Rajper; Rafaqat Hussain Arain; Ghulam Hussain Chandio; Abdul Qadir Bhangwar; Hidayatullah Shaikh; Kashif Hussain Shaikh

The rice crop in the Agriculture field is playing an important role in the economy of Pakistan and fulfilling the needs of the living hood of human beings. The rice leaf faces several diseases like Bacterial Bligh, Brown Spot, Blast, and Tungro. This research attempts to create a simple and best model for rice leaf disease detection using the deep learning model Yolov5. The model has been upgraded to v5 which is the latest version of Yolo. The performance and accuracy of object detection using Yolov5 are better than Yolov3 and Yolov4 models. The envisioned recognition system has numerous applications but here focused on the agriculture field to increase crop productivity by detecting plant diseases at an early stage. This model can differentiate and successfully detect rice leaf diseases. The rice leaf images dataset is downloaded from the Kaggle website, the dataset contains 400 images of leaves infected by a disease. This paper uses the Google collab platform to train, validate and test the model for rice Leaf disease detection. All necessary steps to be implemented, the rice leaf disease are detected and fully described. The developed model utilizes epochs: 100. The experimental results show that the deep learning model created with 100 epochs has shown the best performance with precision, recall, and mAP values of 1.00, 0.94, and 0.62, respectively.