Enhancing Plant Disease Recognition on Mobile Devices: A Hybrid Approach with CNNs, Rule-Based Systems, and Machine Learning

The food industry has significantly boosted the agricultural economy of India, historically known as the largest producing nation with a strong agricultural identity. The primary crops include grains, fruits (such as potatoes, oranges, tomatoes), sugarcane, and specially cultivated grains and cotton. Maharashtra, a state in India, has experienced substantial economic growth, particularly attributed to citrus and cotton industries. This growth has generated employment opportunities and holds significant potential for the state’s economic advancement. To sustain the prosperity of these industries, the government is actively addressing concerns related to disease control, Labor costs, and global market dynamics. In recent times, citrus canker, citrus greening, and black spots on cotton have emerged as severe threats to citrus crops in Maharashtra. Farmers are troubled by the costs associated with tree loss, scouting efforts, and the use of chemicals to control diseases. An automated detection system could play a crucial role in preventing and minimizing these losses, thereby safeguarding the industries, farmers, and the overall economy. This research focuses on developing disease detection through pattern recognition methods for various crops. The image detection process has some main categories which are image acquisition, image processing, and pattern recognition. Image preprocessing is used to process the image in a clear and clean format. Pattern recognition methods are then utilized to classify samples into different crop conditions. To assess classification approaches, results will be compared across different crops for disease detection. The goal is to demonstrate a classification accuracy surpassing existing models, reaching the highest. The research primarily aims to showcase the feasibility of disease detection based on visible symptoms on fruits or leaves. Data collection and initial knowledge acquisition are planned through offline and online approaches. The overall motive is to have accurate detection of the fruits and leaves thereafter using those results for diseases detection.