AN EFFICIENT DEEP LEARNING FRAMEWORK FOR DISEASE DETECTION IN PLANTS

Plant disease remains a persistent challenge for smallholder farmers, posing threats to both income and food.The upgrowth in smartphone emergence which combined with improvements in computer vision models has paved the way for new opportunities in image recognition in agriculture field. Convolutional Neural Networks, a deep learning technique renowned for their excellence in image recognition, offer the potential for quick and accurate diagnoses.we investigate the working of a pretrained Convolutional neural network model in detecting crop diseases.Image processing methods are employed to study images of plant leaves. The developed model is deployed as a user interface (UI) and is capable of recognizing 38 plant diseases from a dataset of more than 87,000 images of healthy and diseased leaf tissue.Rather than detecting disease name we also provide remedy recommendation in our model. We have secured treatment protocols for plant diseases from different sources. Thus, the system provides an indispensable resource for farmers, enabling them to detect diseased plants and take appropriate measures to prevent it.Activation functions in our model utilize Rectified Linear Unit (ReLU) to improve feature extraction efficiency.The dataset, consisting of leaf images captured in a controlled environment, is used for training and validating the model. Results show that the model acquired 94% accuracy after 8 epochs, demonstrating the technical feasibility of Convolutional neural networks in classifying plant diseases and providing solutions farmers.