Plant Disease Recognition

Problem Context

In agriculture and horticulture, visual symptoms such as yellow leaf spots, mold, stem wilt, and discoloration often require experienced plant-protection staff to distinguish among fungal, bacterial, viral, and physiological disorders. When users only have a photo or video, it is hard to quickly determine the likely disease, severity, and next action.

How It Works

The skill accepts image or video input and follows a workflow covering lesion detection, disease classification, severity assessment, and prevention advice generation. It supports local file paths or network URLs for mp4/avi/mov and jpg/jpeg/png, up to 10 MB. During execution, it uses the bundled script scripts/plant_disease_recognition_analysis.py to analyze visible symptoms, environmental context, and crop growth stage, then outputs a structured report containing disease name, pathogenic cause, and control recommendations. Historical report queries use --list to read from the cloud API and render results as a Markdown table, avoiding local memory or manual aggregation.

Boundaries

Captures should show the affected area clearly, preserve typical symptoms, and avoid strong backlighting, overexposure, or fully decayed tissue. The output is a diagnostic reference only and does not replace on-site assessment by plant-protection professionals. The skill does not generate temporary scripts and does not ask users for internal identity parameters.