Computer Vision for ID Scans
Industry
Transport
Client
Kautra
Team
SCALE
Year
2025
We were hired to automate our discount validation and help manage bus crowding – without slowing down operations.
Challenge
Kautra offers discounted fares (students, seniors) that required drivers to manually inspect IDs/cards, slowing down boarding and sometimes causing delays. Popular routes also suffered from overcrowding at peak times, but the company lacked data-driven tools to manage capacity dynamically.
Approach
Developed a mobile ID recognition system using on-board cameras and OCR/vision models to automatically validate discount cards as passengers board. Integrated this with the ticketing app to apply discounts in real time.
Utilized 5G-connected devices on buses to live-stream passenger count data. Built algorithms to detect overcrowding and trigger alerts to dispatch additional buses or adjust timetables.
Implemented a basic dynamic pricing engine that could, in future, offer incentives (lower fares) for off-peak travel or less crowded routes (as a way to redistribute load).
Provided a driver’s tablet interface showing boarding validations and recommendations (e.g. “X seats left, advise standing passengers of next bus in 5 min”).
Architecture/Backend
Microservices/API design with optional cloud offload.
AI/ML models
Computer vision models; algorithms to detect overcrowding; dynamic-pricing engine.
Infrastructure/Deployment
Edge AI on buses using mobile hardware and 5G.
Key results
Boarding process sped up ~20% – automated ID checks shaved ~5–8 seconds off per passenger boarding. Over a busy terminal, this reduced dwell times significantly, keeping buses on schedule.
Overcrowding reduced ~15% on popular routes after pilot implementation. Drivers and operations could proactively deploy extra shuttles when the system forecast demand spikes (e.g. end of university classes).
Positive customer experience feedback: passengers (especially students) saw faster lines and no longer needed to fumble for IDs – the system’s ease was noted in a pilot survey.
