Sefism is free for learners. We are here to help you grow, one step at a time.Join for free
Sefism.
Games & XR

Driving Theory Hazard Perception Game for Learner Drivers on Local Roads

A hazard perception training game for learner drivers built from dashcam clips of local roads, where players tap when a developing hazard appears and are scored on how early they react. A final year project for driving schools in countries whose traffic looks nothing like the UK clips most hazard perception apps use.

Topics
Difficulty

Intermediate

A small team, or one strong student willing to learn something new.
Effort

2 semesters · 3 students

Deliverables

7 to ship

3 optional extras

Suggested stack

FlutterDartFFmpegNode.jsPostgreSQLReact (annotation tool)
A suggestion, not a requirement. Swap anything for what you already know.

What you should ship

  • A clip annotation tool where an instructor marks the start and end of each developing hazard window in a dashcam video and adds a short explanation.
  • A clip pipeline that trims, compresses and packages clips with their hazard windows so a lesson pack downloads once and plays offline.
  • A mobile game that plays clips, records tap timestamps, scores each hazard by how early inside the window the player reacted, and flags pattern tapping as cheating.
  • A review mode that replays the clip with the hazard window highlighted and the player's taps marked on the timeline.
  • At least 30 annotated clips from local roads covering motorbikes, rickshaws, pedestrians crossing, sudden lane changes and livestock, recorded with permission and with number plates and faces blurred.
  • A driving school dashboard showing each student's scores over time and which hazard types they miss most.
  • A study with at least 15 learners comparing reaction scores before and after five practice sessions.

If you have time left

  • Multiple annotators per clip with an agreement score, and automatic use of the median window.
  • A road sign and rules quiz mode in Urdu and English sharing the same progress tracking.
  • Suggesting hazard window candidates with an object detector to speed up annotation.

The problem: hazard training built for someone else's roads

Reading the road is the skill that separates safe new drivers from dangerous ones: noticing the child between parked cars, the motorbike about to cut across, the bus pulling out. Some licensing systems, such as the UK's, test it formally with a hazard perception test built from video clips. Many countries do not, and where learners practise at all, the apps they find are full of UK or Australian roads with lane discipline, zebra crossings and very few motorbikes.

Learners in Lahore, Karachi, Dhaka or Nairobi face a different mix: rickshaws, motorbikes filtering on both sides, pedestrians crossing anywhere, animals, and vehicles going the wrong way. Driving schools teach these hazards verbally or not at all, because they have no practice material.

What already exists and the gap

  • Official DVSA hazard perception material (the UK Official Hazard Perception app from TSO) uses CGI clips scored on how early the candidate responds within each hazard window.
  • Driving Theory 4 in 1 Kit and similar apps bundle theory questions and hazard clips for UK learners.
  • Hazard Perception Test CGI and other commercial apps offer the same format for the UK market.

These apps are well made but their clips are from roads that look nothing like mixed traffic in South Asia or Africa, they are English-only, and none give a local driving school the tools to create and score its own clips. The gap is the tooling plus the local content: an annotation workflow, an offline player and a school dashboard.

What you build

Three parts. The annotation tool (web) lets an instructor upload a dashcam clip, scrub to where a hazard starts developing and where it becomes too late, and write why. The pipeline trims and compresses clips, blurs faces and plates, and packs them into lesson packs. The mobile game downloads packs, plays clips full screen, captures taps with precise timestamps, scores them, and shows a review timeline. The school dashboard groups students and shows progress.

The technically interesting part

Two things make this worth a final year project. First, timing accuracy: a tap must be mapped to the video's presentation timestamp, not the wall clock, or scores drift on slow phones. You need to measure the offset between video playback position and touch events on several devices and correct it. Second, fair scoring: splitting each hazard window into bands and awarding more points for earlier reactions, while detecting rhythmic or excessive tapping, which the UK test treats as an automatic zero for that clip. Document your scoring rules and test them with synthetic tap sequences.

Annotation consistency is the hidden third problem: two instructors will not mark the same window. Measuring that disagreement is a good section for your report.

How to evaluate it

  • Timing accuracy: play a clip with a known flash frame, tap on it with a high-speed recording as ground truth, and report the tap-to-frame error on at least three phones.
  • Scoring validity: unit tests with synthetic tap patterns, including pattern tapping and taps just outside the window.
  • Learning: at least 15 learners take a baseline set, practise five sessions, then take a different set; report the change in mean score.
  • Annotation agreement: two instructors annotate the same 10 clips; report the mean difference in window start times.

Suggested plan

  1. Weeks 1 to 4: partner with one driving school, agree consent and recording rules, record the first clips.
  2. Weeks 5 to 9: annotation tool and clip pipeline with blurring and compression.
  3. Weeks 10 to 14: basic player with tap capture and scoring; timing measurement on devices.
  4. Semester 2, weeks 1 to 5: offline packs, review mode, Urdu text, anti-cheat rules.
  5. Weeks 6 to 9: school dashboard and progress charts; annotate the full clip set.
  6. Weeks 10 to 14: learner study, report and demo.

Path to a product

The first customer is a driving school chain that wants a differentiator and something to give students between lessons. Charge the school per student seat, with a starter pack of clips included and the annotation tool for their own. Later, the same tooling sells to fleet operators and ride-hailing companies that onboard new drivers in bulk. The clip library grows every time a customer annotates, so the product gets better with each school.

Scope warning

Do not build a full theory test app, a driving simulator or a 3D engine. Do not use clips you do not have rights to, and never publish unblurred faces or plates. Automatic hazard detection with deep learning is a stretch goal at most; the product works with human annotation.

Learn what you need on Sefism

Get help with this idea

Most of the difficulty in this project is data: clips, consent and consistent annotation. Bring your proposal and we can agree the scope, pick a stack your team can finish with, and design an evaluation your panel will respect. You build it; we help you make the calls.

New here? Create a free account first.

Ideas and guidance, not finished projects

These are project ideas and scoping guidance, published free for students to use as a starting point. I do not build, write, or sell final-year projects, and I do not complete coursework for anyone. Take an idea, make it yours, and build it.

Not sure this one fits you?

I help students pick an idea that matches their skills and their deadline, then scope it down until it is finishable. Guidance only, never done for you.

Discuss this idea
Work with me

Stuck on something specific?

Writing only gets you so far. If you want an answer to your situation rather than the general case, book a session and we will work through it together. Sessions are free for approved Sefism members, and a few slots open each week.

Book a free sessionSee all services

Follow along

New writing, resources and project ideas land here first.