Skip to content
SIH Buddyby Ganeev Singh
Dev

🔥 Roast My Pick · SIH26124

AI-Powered Mobile Urban Intelligence Platform Using Public Transport Fleet

Bharat Electronics Limited

Medium39/100

Reasonable choice. The scoreboard liked it. The scoreboard is not the one asking questions on the day.

Worth considering. Cut this to road defects plus multi-pass deduplication and say plainly what you excluded — the buses-as-sensors idea is strong, but chasing the full feature list guarantees you lose to a team that did one part properly. Roughly 110–260 teams are expected to go here.

The receipts

Every red flag on this statement, in full. These are the four places it bites.

  1. Exhibit A

    The description asks for road defects, infrastructure gaps, traffic density, pedestrian safety, hit-and-run tracking and OD analytics — attempting all of it produces six shallow features and no depth anywhere

  2. It gets worse

    Plate recognition from a moving bus at speed, in the conditions the description names, is far harder than from a fixed ANPR camera, and it duplicates SIH26127

  3. Still reading?

    Bus-mounted cameras suffer vibration, glare and windscreen reflections that dashcam training data may not capture, so field accuracy will be well below your reported metrics

  4. And the finisher

    Detecting a hit-and-run and identifying an offending vehicle is a law-enforcement capability with evidentiary implications that a confidence-scored detection cannot support

The damage report

Every score this statement earned, and what each one actually costs you.

  • Feasibility

    3/5

    Buildable. Not comfortably. There is a week in here you have not planned for yet.

    Public road damage datasets exist and dashcam footage is easy to obtain, so the core detection is trainable, but the description also demands hit-and-run vehicle tracking with plate extraction, pedestrian risk detection and origin-destination analytics, which is several more products layered on top.

  • Innovation scope

    4/5

    There is something genuinely new here. Do not bury it under another dashboard.

    The multi-pass deduplication problem — recognising that three buses saw the same pothole and fusing those sightings into one confident report — is genuinely open and is the most interesting thing in the statement.

  • Clarity

    3/5

    Clear enough to start, vague enough to drift. Write the scope down and stop reinterpreting it weekly.

    Each individual capability is described precisely, but the description strings together road defects, infrastructure gaps, traffic density, pedestrian safety, hit-and-run tracking and OD analytics without any priority, so a team cannot tell what is actually being graded.

  • Acceptance potential

    3/5

    Middle of the pack. This statement will not win the room for you — you will have to.

    The mobile-sensing concept is genuinely strong and the demo is easy, but the description's breadth guarantees a shallow build, and the ANPR and tracking components overlap SIH26127 from the same organisation so you risk being compared against a team that did only that, properly.

  • Effort

    Massive

    A semester of work wearing a hackathon costume. Something is getting cut; decide what now, not in week five.

    Taken literally this is an edge detection stack, an ANPR system, a tracking system, a geospatial aggregation platform and a traffic analytics engine — five products where the statement implies one.

  • Demo-ability

    Easy

    Easy to demo — and so is everyone else's. Working is the floor here, not the achievement.

    Detections overlaid on dashcam footage and pins appearing on a city map is immediately legible, and real footage from any Indian road makes it feel authentic.

  • Data

    None supplied

    No dataset comes with this one, so every accuracy figure you quote is a number about labels you invented.

    Nothing is provided with the statement. You are sourcing, cleaning and labelling it yourself, and that work is invisible in the demo but very visible in the questions.

The demo they will have already seen

Somewhere around 110–260 teams are heading here, and the description is doing the choosing for most of them. They will read the same brief, reach the same architecture, and build a version of the same demo you are planning. Being correct is the floor. If your five minutes could be swapped with the team before you and nobody in the room would notice, you have not picked badly — you have built predictably, which costs exactly the same and hurts more.

What survives

The ground worth standing on when the questions start.

  • RDD2022 and similar public road damage datasets give you real labelled potholes and cracks from Indian and Asian roads with no collection effort
  • Dashcam footage is trivially available, so you can demonstrate on genuine road conditions rather than staged clips
  • The multi-pass deduplication is a real technical contribution that almost no competing pothole-detection project will attempt

Nothing here is fatal. It is just the list of places this statement pushes back, and you now get to push there first.

Full analysis ↗Roast another

The framing is a joke. The findings are not — they are the same analysis on the statement page, and every line above is attached to a score or a fact in the record. It is one opinion with its reasoning attached, so argue with it before you trust it.