🏞️ FRA Atlas AI — Intelligent Forest Rights Monitoring & Decision Support System

Built for Smart India Hackathon 2025 | Team DevSphere

Digitizing Forest Rights for Transparent, Data-Driven Governance.
FRA Atlas AI automates the verification of forest land rights, integrates geospatial and satellite data, and powers a Decision Support System (DSS) to link FRA patta holders with government schemes.

---

🧠 Problem Statement

Issue	Description
Fragmented Data	FRA claims, GIS maps, and Census records exist separately and are mostly non-digital.
Manual Verification	Field staff verify documents manually without real-time map references.
Lack of Transparency	No centralized system exists to visualize granted titles or scheme eligibility.
Objective	Create an AI-powered, GIS-integrated, end-to-end FRA management platform with a decision-support layer.

---

💡 Our Solution — FRA Atlas AI Platform

Component	Description
OCR Automation	Extracts and structures data from legacy paper FRA documents using Tesseract/Google Vision.
AI Verification Engine	Cleans and validates data via Pandas, Great Expectations, and LLM-based anomaly detection.
Land Classification	Identifies land types (forest, agriculture, pond, etc.) using ISRO Bhuvan API or custom CNN model.
Geo-Mapping	Displays approved claims on an interactive map using Mapbox + PostGIS.
DSS Query Interface	Officers can ask natural language queries (via LangChain + MCP) — e.g., “Show all PM-KISAN eligible families in Village X.”
Citizen Portal	Enables tracking claim status and viewing approved maps (no edit access).

---

🧩 Core Features

Feature	Description
🧾 Document Digitization	Automated OCR and data structuring from scanned FRA documents.
🧹 Data Cleaning Pipeline	Validates extracted text using Pandas, Regex, Great Expectations, and LLM.
🛰️ Land Type Classification	Classifies land via ISRO Bhuvan or a CNN model trained on Sentinel-2 / EuroSAT data.
🗺️ FRA Atlas Visualization	Approved claims plotted on an interactive, zoomable GIS map.
🤖 Decision Support System (DSS)	LLM-powered query engine to layer Central Schemes (PM-KISAN, MGNREGA, etc.) over FRA data.
📬 Smart Notifications	Sends SMS/email to eligible citizens when new schemes are introduced.
👥 Citizen Transparency Portal	Public dashboard showing approved claims (non-confidential info only).

---

⚙️ Tech Stack

Layer	Technology Used
Frontend	React.ts, Redux Toolkit / React Query, Mapbox GL JS, TailwindCSS
Backend	FastAPI, REST APIs, JWT Auth, MCP middleware
AI / ML	OCR (Tesseract/Google Vision), CNN (TensorFlow/Keras), LLM (LangChain + Google-Gemini)
Data Cleaning	Pandas, Great Expectations, Regex
Geo / Spatial	ISRO Bhuvan API, Google Earth Engine, PostGIS
Database	PostgreSQL + PostGIS, Redis (caching), Elasticsearch (search)
Notifications	Twilio / AWS SNS for SMS
Deployment	Docker, NGINX, GitHub Actions CI/CD

---

🏗️ System Architecture

Step	Process
1️⃣	Officer logs in → authentication check.
2️⃣	Uploads scanned FRA document → triggers OCR.
3️⃣	Extracted text cleaned using Pandas + LLM verification.
4️⃣	Land coordinates checked via ISRO Bhuvan / CNN model.
5️⃣	If verified → record stored in PostgreSQL + plotted on GIS map.
6️⃣	DSS allows NL queries like “Families eligible for PM-KISAN in District A.”
7️⃣	Eligible citizens notified automatically by SMS/email.

---

📂 Folder Structure

Folder	Description
Backend/	FastAPI backend, OCR & ML services, API routes, database logic
routers/	REST endpoints for upload, DSS queries, and classification
services/	Scheme overlay service, OCR extraction, and ML inference
Frontend/	React dashboard for officers and citizens
src/pages/	Login, Dashboard, Upload, DSS Query, Citizen Map
src/components/ui/	Header, Chatbot, Layouts, and protected route handlers
models/	CNN model for land classification
fra.db	PostgreSQL database file (with PostGIS)

---

🚀 Quick Start

Step	Command
1️⃣ Clone Repository	git clone https://github.com/jeetgoyal80/FRA-Portal.git
2️⃣ Install Backend Deps	pip install -r requirements.txt
3️⃣ Run Backend Server	uvicorn main:app --reload
4️⃣ Launch Frontend	cd Frontend && npm install && npm run dev
5️⃣ Access in Browser	http://localhost:5173

---

🧠 Model & Data References

Resource	Link
EuroSAT Sentinel-2 Dataset	https://github.com/phelber/EuroSAT
Deep Learning for Land Use & Cover Classification	arXiv:1709.00029
Sentinel-2 Land Cover Classification with CNNs	MDPI Remote Sensing Journal
Forest Rights Act, 2006	IndiaCode PDF
FRA Rules & Guidelines	Ministry of Tribal Affairs
ISRO Bhuvan Portal	https://bhuvan.nrsc.gov.in/ngmaps

---

🎥 Demonstration Video

Type	Link
▶️ Project Demo Video	Watch on YouTube

---

🔒 Privacy & Data Ethics

Principle	Measure
Data Minimization	Only essential details are displayed publicly.
Access Control	Role-based authorization for officers, admins, and citizens.
Transparency	All approved FRA claims visible to public in anonymized form.
Audit Trail	Every action (upload, approval, DSS query) is logged for accountability.

---

🧭 Future Scope

Feature	Description
🌐 Multi-Language NLP	Add Hindi & tribal dialects to DSS query engine.
🛰️ Real-Time Satellite Sync	Live update from ISRO Sentinel datasets.
🤝 Scheme Integration API	Automated scheme mapping for all ministries.
📱 Mobile App	Flutter app for citizen & officer access.
🧮 Predictive DSS	ML-based recommendations for pending approvals.

---

👥 Team DevSphere

Member	Role	Responsibility
Jeet Goyal	LLM, DSS & OCR System Developer	Developed LLM-based DSS engine, integrated OCR automation, LangChain pipeline, and database linking
Rakshit Hinduja	Frontend Developer	Built officer dashboard and interactive map visualization
Harshil Khandelwal	Backend Developer	Designed and implemented REST APIs, FastAPI backend, and database integration
Madhav Gupta	CNN Model Developer	Developed land classification model using ISRO Bhuvan and EuroSAT dataset
Neelam Patidar	Frontend & Research Lead	Assisted in UI/UX development, conducted research, and prepared the final presentation
Vedika Vishwakarma	Presentation Designer	Assisted in designing and finalizing the project PPT

---

🏁 Project Status

Stage	Progress
OCR & Data Cleaning	✅ Completed
Land Classification (CNN)	✅ Completed (EuroSAT dataset)
DSS Query Engine	🧠 Under Development
Frontend Dashboard	⚙️ Under Development
Deployment & Testing	🚀 Upcoming

---

🧾 References

See full list of datasets and legal references in Model & Data References section.

---

© 2025 Team DevSphere | Built for Smart India Hackathon 2025