evidoria

← Back to browse

Good practice Imported

Dhaka's AI-Powered Traffic Violation Detection and E-Prosecution System

Bangladesh · Dhaka · See the Bangladesh profile · See the Dhaka profile

Evidence: Observational / pre–post Top 83% 40/100 · Ask Evidence Copilot about this practice

Dhaka Metropolitan Police deployed AI-enabled CCTV cameras at major intersections to auto-detect traffic violations and generate e-prosecution cases; officials reported roughly 1,000 cases in the first ten days, alongside an adaptive-signal pilot cutting reported wait times.

~200 violations
Violations captured in the camera network's first days (May 2026, first days)
~1,000 cases
E-prosecution cases generated within ten days of launch (7-17 May 2026)
4-5 to 2-3 minutes
Average wait time at pilot intersections, before and after (pilot vs prior)
4.2 km/h
2024 baseline average vehicle speed (2024)
~50,000 USD
Reported cost of the signal-timing pilot (pilot phase)

Details

Maturity
Pilot
Promoter
Dhaka Metropolitan Police, with Bangladesh University of Engineering and Technology (BUET)
Period
2026–present
Keywords
traffic enforcement, law enforcement, transport, public safety

Context

Dhaka Metropolitan Police (DMP) faces chronic traffic-rule violations and congestion, and earlier Dhaka traffic-automation projects since the early 2000s — including a 2016 rollout covering 110 intersections — largely fell into disuse, a relevant caution for this pilot's durability.

Objectives

Automatically detect common traffic violations and generate prosecution cases without manual patrol, and, in a related pilot, adjust signal timing in real time to ease congestion.

Activities

In May 2026, DMP switched on AI-enabled CCTV cameras built with Bangladesh University of Engineering and Technology (BUET) at major intersections including Shahbagh, Karwan Bazar and Bijoy Sarani, detecting red-light jumping, wrong-lane driving, illegal parking, no-helmet riding, mobile-phone use and misuse of VIP lights, and auto-generating e-prosecution cases sent to vehicle owners by SMS and post; flagged footage still passes through manual human review before any penalty is issued. A related pilot with the Dhaka Transport Coordination Authority adjusts traffic-signal timing in real time at seven congested intersections based on camera-derived vehicle density.

Results

Dhaka Tribune reported roughly 200 violations captured in the camera network's first days, and by 17 May 2026 Global Voices cited DMP figures of around 1,000 generated cases. Arab News reported the signal-timing pilot cut average wait times at pilot intersections from 4-5 minutes to 2-3 minutes, against a 2024 baseline average vehicle speed of 4.2 km/h, for a reported pilot cost of about $50,000. DMP has announced plans to add 60 more cameras this year and double that again next year.

Conclusions

All figures are self-reported by DMP via press statements, with counts shifting week to week as the rollout expands (5 to 25 intersections cited across different reports) and no independently audited baseline or accuracy figure for the detection model. Civil-society voices also flag that unregistered vehicles evade detection and that tickets fall on the registered owner regardless of who was driving.

Implementation
Indicative cost
Low (< €50k) — Signal-timing pilot component reported to cost approximately $50,000.
Time to results
Medium (1–3 years) — Launched 7 May 2026; DMP plans to add 60 more cameras within the year and double that again the following year.
Staffing & skills
Dhaka Metropolitan Police (DMP), Bangladesh University of Engineering and Technology (BUET) — camera/AI system development, Dhaka Transport Coordination Authority — signal-timing pilot

Conditions for success

  • AI-enabled CCTV cameras at major intersections
  • Automated e-prosecution linked to vehicle registration and SMS/postal notification
  • Manual human review of flagged footage before penalties are issued

Common failure modes

  • No independently audited baseline or accuracy figure for the detection model
  • Violation/case counts have shifted week to week as intersection coverage expanded (5 to 25 intersections cited across different reports)
  • Three earlier Dhaka traffic-automation projects since the early 2000s (including a 2016 rollout of 110 intersections) largely fell into disuse
  • Unregistered vehicles evade detection; tickets fall on the registered owner regardless of who was driving

Where it fits

Governance type
municipal police force with university partnership
Scale
city-wide pilot, expanding
Income level
lower-middle-income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful