AI-enabled CCTV trial at 14 construction sites as S’pore records 36 workplace deaths in 2025

1 month ago 93

SINGAPORE - At a construction site in Pasir Ris, safety supervisors do not need to constantly monitor screens to keep track of workers’ safety.

Instead, their phones receive instant notifications of safety violations whenever closed-circuit television (CCTV) cameras augmented with artificial intelligence detect unsafe practices.

The SafeSite VA system has been trialled at 14 worksites being managed across seven public sector agencies since the second half of 2025. The trial will continue until June before a decision is made on possibly scaling it to other worksites, said Minister of State for Manpower Dinesh Vasu Dash on Feb 10.

It is an initiative by the Ministry of Manpower’s (MOM) Workplace Safety and Health Institute, the Government Technology Agency of Singapore, and the Prime Minister’s Office’s Public Sector Science and Technology Policy and Plans Office.

During a site visit to Pasir Ris Coast Industrial Park on Feb 10, the system was showcased to the media at the construction site of Riviera station, which will connect to the Punggol LRT line as part of the Cross Island Line Punggol extension.

That worksite, which measures 6ha, is equipped with 31 CCTV cameras, of which 10 are equipped with the video analytics system.

When an unsafe practice is detected, it immediately alerts safety supervisors through Telegram, providing information such as the time of occurrence and the camera that caught it.

This allows supervisors to respond quickly and address the situation on the ground.

Currently, the AI system flags workers who fail to wear personal protective equipment, stand under suspended loads, near machinery or vehicles, or work at a height in areas without railings.

It also has geofencing capabilities – workers detected in unauthorised areas will also be flagged.

Mr Dinesh said: “When the cameras are around, that becomes the permanent safety officer that’s constantly watching, and that allo...

Read Entire Article