# Implementation Epics

## EPIC 01 — Repository and Runtime Foundation

Goal: create a runnable full-stack foundation with backend, frontend, database, worker and shared conventions.

Must deliver:

- FastAPI app booting
- health endpoints
- SQLAlchemy/Alembic setup
- PostGIS-enabled PostgreSQL connection
- React app booting
- Docker Compose local stack
- basic CI-friendly test commands

## EPIC 02 — Project and Area Management

Goal: allow users to create project workspaces and define Kempen analysis areas.

Must deliver:

- project CRUD
- area CRUD
- polygon validation
- area surface calculation
- geometry storage in PostGIS
- frontend project dashboard
- map drawing workflow

## EPIC 03 — Dataset Manager

Goal: upload, register and inspect geospatial datasets.

Must deliver:

- dataset upload endpoint
- metadata extraction
- storage path registration
- dataset table
- dataset detail page
- processing status model

## EPIC 04 — Raster Foundation

Goal: support GeoTIFF metadata, preview, clipping and tiling readiness.

Must deliver:

- Rasterio metadata extraction
- bounds, CRS, band count, resolution
- clip request contract
- tile generation contract
- raster preview placeholder backed by real metadata

## EPIC 05 — Vector Foundation

Goal: support GeoJSON/Shapefile/GPKG import and core vector operations.

Must deliver:

- GeoPandas import
- CRS transform
- clip to area
- area/perimeter calculation
- vector layer display/export

## EPIC 06 — GRB Reference Integration

Goal: retrieve GRB features for selected areas and cache them in PostGIS.

Must deliver:

- GRB source config
- WFS request service
- cache table or dataset registration
- building reference layer
- failure handling and source metadata

## EPIC 07 — Detection Pipeline Skeleton

Goal: run object detection over prepared raster tiles and store georeferenced outputs.

Must deliver:

- model registry stub
- detection job request
- tile inference interface
- detections table
- GeoJSON export
- map overlay

The first implementation may use a deterministic fixture model only if the full interface is identical to a future YOLO implementation.

## EPIC 08 — QA/QC Engine

Goal: compare detections against GRB reference features.

Must deliver:

- spatial matching
- IoU calculation
- precision/recall/F1
- false positives/false negatives
- QA result layer
- metrics panel

## EPIC 09 — Export Layer

Goal: export useful outputs.

Must deliver:

- GeoJSON export
- analysis metadata export
- QA summary JSON
- future-compatible report registry

## EPIC 10 — Stabilization and Polish

Goal: make the MVP coherent, testable and demo-ready.

Must deliver:

- error states
- empty states
- loading states
- validation messages
- test coverage on critical services
- changelog and release notes