Agriscan

Agriscan is a web-based application developed to detect diseases in subtropical fruit plants using deep learning and image processing techniques. This platform aims to assist farmers and agricultural stakeholders in identifying plant diseases early, thereby reducing crop losses and improving agricultural productivity. By simply uploading an image of a fruit plant leaf, users can receive fast and accurate disease classification along with recommendations for treatment. The core functionality of Agriscan relies on a convolutional neural network (CNN) model that has been trained to distinguish between healthy and diseased plant conditions.

To build this system, various Python libraries were employed. The dataset was sourced from Kaggle and managed using libraries such as shutil, os, and zipfile for file organization and extraction. Images were preprocessed using cv2 (OpenCV) and the Image module from PIL to ensure consistent input formats. Numerical computations were handled with numpy, while matplotlib was used to visualize training progress and prediction outcomes. The deep learning model was constructed using TensorFlow’s Keras API, including ImageDataGenerator for augmentation and modules like layers and models for building the CNN architecture. Additionally, the json library was used for configuration and metadata management. Together, these tools support Agriscan in delivering a practical and intelligent solution for plant disease detection.