Temporal change detection on satellite image

Temporal Change Detection — Prototype

1) AOI Selection

Draw a rectangle on the map to set area of interest. You can also paste bounding coordinates.

Or paste bbox (lat1,lng1,lat2,lng2)

2) Load Two Images (T0 & T1)

Images must be georeferenced to the same geographic bounding box. For demonstration upload PNG/JPEG and provide TL/BR coords below.

T0 image (earlier) T1 image (later)

Top-left lat,lng

Bottom-right lat,lng

3) Compute Change

Simple pixel-difference (RGB) -> grayscale -> threshold mask. For production, replace with model-backed change classifier.

Threshold (0-255)

4) Timeline & Visualization

Slide to fade between T0 and T1, view change heatmap overlay.

Notes

• This is a front-end prototype: for accurate, production-level change detection use co-registered multispectral stacks and AI models (Siamese nets, CNN+CRFs).
• Server-side options: run Detectron2/UNet models on tiles and return change masks; integrate with GEE or SentinelHub for auto-ingest.