AI-powered drones to help detect hill fires during Chung Yeung Festival

Hong Kong’s Fire Services Department will launch three new AI-assisted drone technologies to strengthen fire detection and its response to heightened risks during tomb-sweeping activities ahead of the Chung Yeung Festival on Sunday.

Wong Tak-fai, acting divisional officer for information and communications technology, said on Friday that the technology would improve the speed and accuracy of fire detection to help the department operate more efficiently.

“Artificial intelligence [AI] not only speeds up the processing of drone images but, more importantly, helps our staff continuously analyse large amounts of images from different drones across various locations to identify situations needing priority attention,” Wong said.

With the Chung Yeung Festival approaching, the department will strengthen manpower to monitor high-risk locations for fires and step up district patrols and public education efforts with the Civil Aid Service and other district organisations.

The department will also deploy 159 firefighters on Saturday across 12 locations in New Territories North, 10 spots in Kowloon and another six locations in New Territories South. On Sunday, an additional 166 firefighters will be deployed.

Hong Kong recorded 693 hillfires in 2025.

The department will also mobilise a newly bought remotely operated aerial surveillance system consisting of an off-road vehicle from mainland Chinese manufacturer Geely retrofitted with two drone docking systems and a generator.

The HK$1.2 million system will allow officers to survey hard-to-reach fire hazard areas. Its drones can conduct automated patrols within a 3km radius and be controlled remotely.

A real-time artificial intelligence video analysis platform will also be used to speed up the analysis of drone footage and images to generate timely alerts for fire hazards or drownings.

The system uses AI to accurately identify signs of smoke and people who may be drowning, reducing the need for staff to review drone surveillance footage.

“The new platform will be able to receive the footage in real time during a drone flight and then simultaneously run an artificial intelligence analysis,” Wong said.

Currently, footage analysis takes place after a flight ends, with officers manually transferring the files from an SD card in the drone to a computer.

The department will also have two high-payload drones equipped with water-bomb systems ready for deployment against remote hill fires. Each drone costs HK$500,000, including batteries.

These drones can carry weights up to 150kg and are equipped with an 80-litre water tank, allowing them to release water bombs where needed.

Wong said that in departmental tests, while it took firefighters one hour to carry rescue equipment 5km uphill, the large drone could do it in 10 minutes.

“The Fire Services Department is exploring the use of large drones in frontline operations, hoping to move from surveillance and observation to delivering firefighting equipment,” Wong said.

However, the acting divisional officer said artificial intelligence would in no way replace firefighters, as he stressed that all operational decisions would still be made by humans.