To get good yield and production in agriculture, there is a need for diagnosing diseases in plants at an earlier stage. For humans it is difficult to detect a particular type of disease. Advanced Machine Learning and Deep Learning algorithms are proficient for detecting and distinguishing the type of disease in plants. In this paper we used Convolutional neural networks and self-designed image segmentation technique symptom threshold to detect the disease in jowar plant, models are optimized using adaptive learning mechanism and regularized to overcome overfitting. The main aim of this research is to diagnose Anthracnose and Leaf Blight in Jowar plant using self-designed and predefined ResNet50 Convolutional Neural Network (CNN) models, back then preprocessing image using self-designed Symptoms Threshold segmentation technique. Model attained 97 percent accuracy in predicting diseases in jowar plant.
Data/CompressedSegemneted4 consist of images to train model
Folder consist of All ipynb files trainng the model files
saved model can used for predictions
Model failed on testing data on production data attaining 52% accuracy