Service Needs

Reference No.

N-0012

Title

Automatic Traffic Survey for Pedestrian and Vehicular Traffic

Business Needs / Challenges

Currently, traffic count surveys of vehicles and pedestrians at road junctions and pedestrian crossings are mainly performed by manual counting and recording on site.  Processing of the collected data entail massive manual efforts and are prone to human error.   Automatic video analytic technologies for pedestrian/vehicle counting and classification are considered suitable to improve the efficiency and accuracy of the traffic surveys.  Proof of concept at roads junctions in Hong Kong including roundabout, signalized road junction with pedestrian crossings, to verify the feasibility and accuracy of the proposed solution under Hong Kong conditions is considered necessary.

Application Areas

Transport

Expected Outcomes

1) there will be no insurmountable problems in mounting and powering the equipment at road junctions.
2) adverse lighting and weather conditions can be mitigated.
3) accuracy in pedestrian and vehicle counting is better than 90%.
4) classification of vehicles into ten classes according to Annual Traffic Census.
5) privacy impact assessment will be carried out and this can confirm that the Personal Data (Privacy) Ordinance is complied with.
6) estimated cost and programme of full implementation will be recommended.

Technologies to be Used

Artificial Intelligence (AI)
Deep Learning
Machine Learning
Video Analytics