Researchers combined deep learning, grey wolf optimization, explainable heatmaps and a Gemini-powered chatbot to detect finger millet diseases with 98.36 percent accuracy.