Skip to content
SIH Buddyby Ganeev Singh
Dev
All problem statements
SIH26127Worth consideringacceptance 3/5

City-Wide AI Engine for Multi-Camera ANPR Trajectory Tracking and Urban Traffic Analytics

Bharat Electronics Limited · Transportation & Logistics · Software

The trajectory view is the memorable part, so build that and the fuzzy plate matching that makes it robust — but be careful about the ninety percent claim, because stating a number invites being tested against it.

What it actually is

Cities have hundreds of number-plate cameras but each one works in isolation, so nobody can follow a single vehicle across the network or extract city-wide movement patterns. The ask is an engine that reads plates accurately in bad conditions, reconstructs any vehicle's full journey across cameras, and turns the aggregate into traffic analytics. The description sets a specific accuracy bar.

What to build

A three-part platform: a plate recognition engine reading Indian number plates across varying light, weather, viewing angle, motion blur and physical plate damage, reporting a per-read confidence; a trajectory reconstruction module that links reads of the same plate across geographically distributed cameras into an ordered journey with timestamps, direction and inferred route drawn on a GIS map; and an analytics layer aggregating all reads into traffic density measures, origin-destination matrices, congestion bottleneck identification and live movement heat maps for the city authority.

Smallest thing that wins the room

Search a specific number plate and watch its complete journey reconstruct across six camera locations on the city map in time order, with the gaps between cameras interpolated and each read's source frame and confidence shown alongside.

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.

Busy180–410 teams expectedroughly 1 in 149–344 wins it

Quieter than 20% of the 226 · #180 of 226 by expected field

Busier than most. Expect several teams to arrive at the same obvious solution.

Why: company-sponsored statements drew the smallest fields of all.

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

  • YOLOv8 plate detection + PaddleOCR / TrOCR recognition
  • Indian plate datasets (IDD, ALPR benchmarks)
  • Plate-string fuzzy matching for partial reads
  • PostGIS trajectory reconstruction and map matching
  • Kepler.gl / deck.gl OD flow and heat maps
  • Kafka multi-camera stream ingestion
  • Automatic number plate recognition
  • Urban surveillance analytics
  • Spatiotemporal tracking

Prior art to read before you start

multi-camera vehicle trajectory tracking · plate recognition under adverse conditions · origin-destination traffic analytics

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.