mint-1

byDivya Rajput

1 Project Setup 2 React Dashboard 3 FastAPI Backend 4 GeoTIFF Validation 5 Planner Agent 6 AI Models 7 GeoReason Engine 8 Dashboard + PDF Phase 1 — Project Setup Ye sabse pehle banta hai. Languages Frontend: TypeScript Backend: Python AI Tools Work Tool Code ChatGPT UI Figma Git GitHub Deployment Docker Folder: satquery-x/ │ ├── frontend/ ├── backend/ ├── ai_models/ ├── database/ └── datasets/ Phase 2 — Dashboard (React) Sabse pehle UI banao. 6 Tech React Next.js Tailwind Pages Page Purpose Home Landing Upload Images Dashboard Results Report PDF Components: Navbar Upload Zone Image Preview Chat Box Map Confidence Card Phase 3 — Backend React directly AI se baat nahi karega. Beech me FastAPI hoga. APIs POST /upload POST /validate POST /analyze GET /report Language: Python Framework: FastAPI Phase 4 — GeoTIFF Validation User image upload karega. Python me Rasterio metadata read karega. Validation Flow Read GeoTIFF Read CRS Check Resolution Detect Sensor Libraries: Rasterio GDAL Output: CRS ✓ Resolution ✓ Sensor = Optical Phase 5 — Planner Agent (Required by ISRO) Planner sirf routing karta hai. Example: User Query Highlight flooded roads. Planner Output Task = SAR Flood Analysis Models = Grounding + SAR Fusion Input = Optical + SAR Pair Tech: Python Llama 3.1 Prompt Engineering Planner kuch train nahi karega. Phase 6 — AI Models Ye alag-alag specialist models hain. 6.1 VQA Question: Describe this image. Model: RemoteCLIP RSVQA Output: River Buildings Forest 6.2 Grounding Question: Highlight water body. Model: GroundingDINO Output: Blue polygon 6.3 Change Detection Input: 2023 + 2026 Model: ChangeCLIP Output: Urban growth Forest loss Roads 6.4 Optical + SAR Purpose: Flood detection Model: SEN12MS encoder Output: Verified flooded villages Phase 7 — GeoReason Engine (Innovation) Ye hamara original module hai. Ye verify karega: Optical SAR Metadata Temporal Final confidence generate karega. Language: Python Phase 8 — Database Store karega: Table Data Users Login Projects Analysis Images Metadata Reports PDF Database: PostgreSQL + PostGIS Phase 9 — Final Dashboard 6 User dekhega: Left Satellite map Highlighted objects Right AI answer Confidence Objects Bottom Execution summary Download report Phase 10 — PDF Report Python: ReportLab PDF pages: Cover User Query Images Highlighted Map Statistics Confidence Models Used Complete Tech Stack Module Language Library UI TypeScript React + Next.js Styling CSS Tailwind Backend Python FastAPI AI Python PyTorch VQA Python RSVQA Grounding Python GroundingDINO Change Python ChangeCLIP GIS Python Rasterio + GDAL Database SQL PostgreSQL PDF Python ReportLab AI Tools to Use Purpose AI Tool UI Code ChatGPT Backend Code ChatGPT Auto-complete GitHub Copilot UI Design Figma AI Testing Postman Version Control GitHub Team Division Member Work AI-1 VQA + Caption AI-2 Change Detection AI-3 Optical–SAR Fusion Backend FastAPI + Planner Frontend React Dashboard ML Ops Database + Docker 24-Hour SIH Execution Hours Task 0–2 Setup & Git clone 2–5 React UI 5–8 FastAPI APIs 8–12 AI model integration 12–16 Planner + Validation 16–20 GeoReason + Dashboard 20–24 Testing + PPT + Demo Final Innovation (Judges ko ye bolna) Planner Agent ISRO ka required feature hai. Hamari originality GeoReason Engine hai, jo Optical, SAR aur temporal outputs ko cross-verify karke explainable confidence score aur contradiction-free final answer generate karta hai. Ye hi SATQUERY X ki unique contribution hai.