RAISE

Rice is the most important crop in India. Biotic and abiotic stresses always play a crucial role in limiting rice production. Farmers frequently encounter challenges in accurately identifying stress symptoms due to their overlapping and complex nature. The RAISE – Rice AI Stress Evaluator project seeks to leverage recent advances in artificial intelligence (AI), machine learning (ML), deep learning, and mobile technologies to develop an AI-driven, image-based diagnostic application that delivers instant, real-time advisory services through simple image capture.
A robust machine learning classification pipeline was developed, incorporating image preprocessing and color-based feature extraction techniques. An ensemble AI model was implemented using TensorFlow and Keras, achieving approximately 99% accuracy on the training dataset and 85–90% accuracy when validated with field-acquired images.
An integrated mobile application has been designed to seamlessly deploy the AI models and associated APIs. The application can diagnose a broad spectrum of rice crop stresses, including major insect pests, diseases, nematodes, weeds, salinity, drought, and nutrient deficiencies. Furthermore, the platform supports the addition of new stress categories, image annotation, and automated model retraining, thereby ensuring scalability, adaptability, and sustainability.