Skip to content
SIH Buddyby Ganeev Singh
Dev
All problem statements
SIH26012Strong pickacceptance 4/5

AI-Based Automated Urban Parcel Mapping and Cadastral Feature Extraction System using Drone lmagery

Ministry of Rural Development · Robotics and Drones · Software

The rare land-records statement with genuinely available training data and a demo that explains itself, but the differentiator is clean parcel topology rather than segmentation accuracy, so budget most of your time downstream of the model.

What it actually is

Turning drone photographs of a town into a property map is currently done by someone tracing boundaries by hand, which takes months. The ask is software that looks at the drone imagery and draws the plot boundaries, buildings and roads automatically. A human surveyor should then be able to correct the result in a web map.

What to build

An automated cadastral extraction pipeline taking orthorectified drone imagery with DSM and DTM as input, running segmentation models to produce four output classes the description names — parcel boundaries, building footprints, roads and access corridors, and land-use classification — then vectorising those masks into topologically valid parcel polygons with shared edges rather than overlapping blobs, running a validation pass that flags overlapping or inconsistent geometries and gaps against existing GIS parcel layers, and delivering it through a Web-GIS interface where a surveyor reviews each extracted parcel, edits vertices, accepts or rejects, and exports GIS-ready cadastral output.

Smallest thing that wins the room

Drop a raw drone orthophoto tile of a dense settlement onto the map and watch parcel polygons and building footprints appear over it in seconds, then drag a vertex and show the shared boundary of the neighbouring parcel move with it.

How crowded this one gets

A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.

Moderate110–260 teams expectedroughly 1 in 96–221 wins it

Quieter than 60% of the 226 · #91 of 226 by expected field

A normal-sized field. Your idea has to be good, not miraculous.

Why: central ministry statements sat below the average.

This is a guess, not a fact

Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.

The scores

The number is the shorthand. The line under it is the reason.

What you will be writing

  • SAM / U-Net / Mask R-CNN aerial segmentation
  • GDAL + rasterio orthoimagery processing
  • shapely + topojson polygonisation with shared edges
  • PostGIS topology validation
  • GeoServer + OpenLayers Web-GIS editor
  • DSM/DTM normalised height differencing
  • Geospatial AI
  • Urban land records
  • Computer vision

Prior art to read before you start

building footprint extraction · automated parcel boundary delineation · Web-GIS cadastral editing

Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.