Early detection and management of crop diseases and pest infestations
“Farmers often lose crops because they spot diseases too late or use the wrong chemicals. This project asks for a mobile-first system that helps farmers identify crop issues instantly via photos, predicts outbreaks using weather data, and connects them with government experts for the right advice.”
Deconstructed Problem Blueprint
WHAT TO BUILD?
A comprehensive AI-driven decision support system consisting of a mobile app for farmers, a web dashboard for government extension workers, and a backend predictive engine.
WHO IS IT FOR?
Primary: Small-scale farmers. Secondary: Agriculture extension officers, government officials, and local agricultural labs.
WHY DOES IT EXIST?
Lack of real-time expert access, delayed manual diagnosis, and reliance on trial-and-error chemical usage leading to soil degradation and financial loss.
INPUT → PROCESS → OUTPUT ARCHITECTURE PIPELINE
Smartphone camera images, GPS coordinates, local weather API data, soil sensor readings, and historical pest outbreak logs.
Computer vision for symptom classification, time-series forecasting for risk assessment, geospatial clustering for hotspot mapping, and RAG-based LLM for multilingual advisory generation.
Real-time disease alerts, localized treatment recommendations, GIS-based heatmaps for officials, and automated SMS/WhatsApp notifications.
CONSTRAINTS
- Must work in low-bandwidth/offline-first scenarios
- Support for Marathi and other regional languages
- High accuracy in variable lighting/field conditions
- Compliance with government data privacy standards
JURY METRICS
- Model inference accuracy (F1-score)
- System response time (latency)
- User interface simplicity (NPS/SUS score)
- Geospatial mapping precision
CORE BOTTLENECK
The 'Data Scarcity vs. Diversity' gap: Collecting high-quality, labeled images of specific regional crop diseases under varying field conditions (lighting, shadows, blur) is significantly harder than building the app UI.
Why You Should Choose This PS (Pros)
- High social impact with clear alignment to government goals
- Strong potential for scaling into a national-level agricultural tool
- Clear, modular deliverables that allow for rapid prototyping
Challenges & Risks to Watch (Cons)
- Requires robust domain knowledge in plant pathology
- Risk of 'over-engineering' the AI while ignoring the farmer's actual UI needs
- Dependency on external weather/satellite APIs