New smart video system on roads can help identify cause of traffic accidents, detect illegal parking

SINGAPORE – Plans are under way to roll out a new intelligent road video platform that can detect road hazards and illegal parking, and also find out what caused a traffic accident.

The new system will gather video footage of, for instance, vehicles that have stopped unexpectedly on the roads, vehicles travelling against the direction of traffic, and pedestrians or cyclists on expressways or in tunnels.

It will also help alert the authorities to objects blocking traffic lanes, including parts of vehicles or construction debris, fallen tree branches or animal carcasses, and hazardous materials.

This new platform will be rolled out in stages from 2028, said the Land Transport Authority (LTA) in response to queries.

A key feature of the system will be the ability to alert the authorities automatically when such incidents occur, according to tender documents put up on government procurement site GeBiz on Oct 16 and seen by The Straits Times.

LTA said it is developing this platform to enhance its existing traffic monitoring capabilities.

The upgraded system provides a more comprehensive view of traffic conditions, which allows quicker identification of traffic disruptions and better management of traffic, it added.

CCTV camera footage from arterial roads, expressways and road tunnels will be consolidated on this platform, while video analytics will be used to detect traffic incidents and collect data, said LTA.

As part of the project, LTA said existing CCTV cameras will be replaced and greater camera coverage will be added at specific locations, where needed, to strengthen traffic monitoring.

According to the tender documents, the platform is also expected to have the ability to detect fire or smoke that affects the safety of road users, as well as flash floods or water ponding in tunnels.

It should be able to assess video footage and come up with a chronological timeline and possible causes of an incident.

LTA added that this system will have the ability to recognise recurring incident patterns and predict potential problem areas, on top of highlighting incident hotspots.

It will also need to support, at the same time, the processing of at least 960 video streams from cameras placed at junctions and arterial roads all over the island.

The platform will then evaluate how serious an incident is, and how badly traffic flow is affected.

Capable of counting and recording traffic volume data over various time periods, it can measure and record average traffic speeds of vehicles.

It can also classify vehicles into categories including cars, taxis, vans, light goods and heavy goods vehicles, public and private buses, as well as motorcycles.

The tender for the new platform is expected to be awarded in the second quarter of 2027.

LTA is also exploring the idea of detecting instances of illegal parking within specific areas in Singapore.

Offences include stopping at a zebra crossing or along a road with a single white zig-zag line, parking within 6m of a junction, not complying when there is a “no waiting” sign, and stopping a vehicle in a direction against the traffic flow.

LTA may exercise the option in this tender to expand the backend system infrastructure for illegal parking cameras, such that it can process at least 2,500 illegal parking camera video streams.

To detect illegal parking, the system needs to capture two images of each illegally parked vehicle, with one clearly showing the vehicle licence plate number, and another contextual image showing the vehicle and its surroundings.

In each instance, the type of vehicle, its colour, make and model – in addition to the date, time and duration of the offence as well as the road name – will have to be ascertained and recorded.

However, a staff member has to review the scenario before confirming that the detected offence is valid.

LTA is also looking into the extension of the backend system infrastructure for road facility construction cameras.

The improved system has to process at least 200 camera video streams to detect non-compliant safety incidents in work zones.

Unsafe work practices include instances where workers do not have proper personal protective equipment, use of unauthorised vehicles in work zones, lack of barricades or traffic cones, workers standing too close to heavy machinery and water ponding in work zones.