Skip to main content
AboutProblemsGuidelinesTerms & conditions

Problem Statement 7

"Mitti-Scan" - Soil Health Card Digitizer + Actionizer​

Focus: OCR (Optical Character Recognition) | Agricultural Chemistry | Product Recommendations | Marketplace Integration

1. Background:​

Government has distributed 25+ crore Soil Health Cards to farmers. These cards contain critical NPK values (Nitrogen: 140, Phosphorus: 12, pH: 5.5), but farmers can't interpret the numbers. Data remains on paper, doesn't translate to action or better yields.

2. Current Problem:​

1. Chemical Jargon

  • Numbers like "N: 140, P: 12, K: 180" are meaningless to farmers
  • No context on what's low/medium/high
  • Impact: Cards sit unused in drawers

2. No Actionable Advice

  • Card doesn't say which fertilizer to buy
  • Can't connect deficiency to market products
  • Impact: Farmers buy wrong inputs, waste money

3. Manual Data Entry

  • Government data stays on paper, not digitized
  • Can't track soil health trends over years
  • Impact: Miss patterns in soil degradation

3. Goal:​

Build an app that converts paper Soil Health Cards into shopping lists with fertilizer recommendations.

Beyond digitization → Create a platform that:

  1. Scans paper cards using OCR
  2. Interprets chemistry values for crop-specific needs
  3. Recommends specific fertilizer products
  4. Calculates quantities and costs per farm size

4. Expected Solution (MVP Requirements):​

1. Scanner Module (OCR)​

Digitize paper cards:

  • Image Upload: Camera or gallery photo
  • OCR Engine: Tesseract / Google Cloud Vision / Azure Vision
  • Data Extracted: N, P, K, OC (Organic Carbon), pH values
  • Table Recognition: Parse standard Soil Health Card format

Output: Structured JSON {N: 140, P: 12, K: 180, pH: 5.5}

2. Verify & Edit Screen​

Quality control:

  • Display Detected Values: Show OCR results in editable form
  • Manual Corrections: Allow farmers to fix misreads
  • Crop Selection: Choose target crop (Wheat, Rice, Cotton)

Demo Goal: Upload card → See extracted values → Edit if needed in 30 seconds

3. Logic Engine (Deficiency Analysis)​

Analyze soil health:

  • Ideal Standards: Compare against crop-specific requirements (Wheat needs N greater than 150)
  • Deficiency Detection: Flag "Nitrogen is Low by 20%"
  • Priority Ranking: Order deficiencies by severity

Output: List of {nutrient, status, deficiency_percentage}

4. Marketplace Recommendations​

Generate shopping advice:

  • Product Mapping:
    • Low Nitrogen → "Neem Coated Urea"
    • Low pH (Acidic) → "Agricultural Lime"
    • Low Phosphorus → "Single Super Phosphate (SSP)"
  • Product Cards: Display with images, approximate prices
  • Retailer Links: Connect to local agri-shops or e-commerce

Validation: Test with 5 real Soil Health Cards, verify recommendations match agricultural guidelines

5. Impact Dashboard​

Real-time visualizations:

  • Health Speedometers: Red/Yellow/Green gauges for N, P, K
  • Deficiency List: Prioritized issues with severity
  • Shopping Cart: Recommended products with quantities
  • Impact Metrics:
    • Cards scanned
    • Deficiencies identified
    • Estimated cost of corrections

5. "Level Up" Features:​

Advanced Features (Choose 2-3)

  • Cost Calculator: Input farm size → Get exact bags needed + total cost
  • Visual Graphs: Bar charts comparing current vs ideal values
  • Trend Tracking: Compare cards from Year 1 vs Year 2
  • Organic Alternatives: Suggest vermicompost instead of chemical fertilizers
  • Crop Yield Predictor: Estimate yield improvement after corrections

6. Tech Stack:​

Core

  • Backend: Node.js/Express
  • Database: MongoDB (scan history, recommendations)

AI/ML

  • OCR: Google Cloud Vision API / Tesseract.js
  • Text Parsing: Regex for extracting numbers

Frontend

  • Framework: React.js
  • Camera: React Webcam / Native camera API
  • Visualization: Recharts (gauges, bars)

External APIs

  • Fertilizer product databases, agricultural reference standards (ICAR guidelines)