Skip to content
SIH Buddyby Ganeev Singh
Dev
All problem statements
SIH26175Strong pickacceptance 4/5

DepthWizard - Single-View Height Estimation and 3D Flythrough

Indian Space Research Organisation(ISRO) · Miscellaneous · Software

Well-scoped with free data and a memorable flythrough demo, and strong depth backbones give you a real starting point — invest in the metric anchoring and be honest about single-view height accuracy, because that is exactly where an ISRO judge will check your metres.

Data: Any open high-resolution remote-sensing dataset

What it actually is

Elevation models normally come from stereo pairs, LiDAR or radar, which are costly and sensor-dependent. Estimating height from a single image is cheaper but monocular depth models predict relative depth and are trained on ordinary photos, not satellite views. The ask is a pipeline that turns a single overhead RGB image into an elevation map — relative for plain images, absolute metric height for georeferenced ones — and lets you fly through it in 3D.

What to build

An end-to-end pipeline that takes a single overhead RGB image and produces an elevation surface: for a plain non-georeferenced PNG or JPG, a relative digital surface model; for a georeferenced GeoTIFF, an absolute metric DSM, obtained by using a pre-trained monocular depth backbone for the initial relative depth and then anchoring it to metric elevation using a coarse DEM source like SRTM or a few ground control points to resolve absolute scale, followed by a real-time interactive 3D flythrough of the reconstructed surface for visualisation and navigation.

Smallest thing that wins the room

Feed a georeferenced satellite image of an urban area, produce an absolute DSM anchored to SRTM, show building heights reading in metres against known references, then launch the interactive 3D flythrough over the reconstructed terrain and structures.

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.

Moderate150–340 teams expectedroughly 1 in 125–289 wins it

Quieter than 34% of the 226 · #150 of 226 by expected field

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

Why: defence, intelligence and space bodies drew small fields.

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

  • Monocular depth backbone (Depth Anything / MiDaS)
  • SRTM DEM metric anchoring
  • Relative-to-absolute scale fusion
  • GeoTIFF georeferencing (rasterio / GDAL)
  • DSM generation
  • Real-time 3D flythrough (three.js / CesiumJS)
  • Single-view depth estimation
  • Digital surface models
  • 3D visualisation

Prior art to read before you start

monocular height estimation from satellite · relative-to-metric depth anchoring · interactive 3D terrain flythrough

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.