Grand Challenge 2022 winners develop AI fire detection system

Team RegeX from the University of Colombo School of Computing in Sri Lanka developed a low-cost early-warning fire detection prototype that uses machine learning and artificial intelligence (AI) and integrates with existing CCTV systems.

Written by Rei Fortes

Fire is one of the many issues that cause severe damages and casualties in densely populated cities today. Rapid urbanisation continues to influence the construction of buildings in crowded metropolitan areas, increasing the need to ensure adequate fire safety systems are in use to prevent fire outbreaks.

A future-ready solution to aid in detecting fires early is in place thanks to the next generation of computer science and information technology undergraduate students from the winning team of La Trobe University’s 2022 Technology Infusion Grand Challenge competition.

Team RegeX from the University of Colombo School of Computing in Sri Lanka developed an automatic real-time early-warning fire detection system called FiremeX, that can be integrated with existing CCTV systems capable of identifying all types of fires using AI and machine learning.

FiremeX is a smart all-in-one system design that can detect both indoor and outdoor fires by integrating with existing digital video record (DVR) or network video recorder (NVR) units and adding a fire detection software that can accurately detect smoke and early signs of a fire combustion.

“Our aim was to create a system that uses machine learning and advanced sensors to detect fires early, allowing for timely responses and potentially saving lives and property,” said Dilan Perera, a member of Team RegeX.

“We also wanted a reliable and affordable solution for businesses and homeowners to integrate FiremeX with their existing systems making it more accessible for everyone.”

Once a fire is detected by FiremeX, the system will automatically notify emergency contacts added by the user who can respond promptly to extinguish the fire or call on firefighting services. Extensive hours of testing were also conducted to reduce the possibility of issues and ensure accurate detection of fires.

“We increased the dataset that we used to train the algorithm by including more images of real-life fires. This helped fine-tune the algorithm and significantly reduce false alarms,” said Thenuka Ovin, a member of Team RegeX.

“Adjusting the CCTV camera settings to reduce the delay of the video stream also helped optimise the algorithm.”

The prize for the team of five undergraduate students is a two-week all-expense paid mentorship stay with La Trobe University’s Centre for Technology Infusion (CTI) in Melbourne, Australia. Team RegeX will have the opportunity to learn from academics and business experts from across La Trobe University.

“We’re thrilled to have won the competition and are eagerly looking forward to our mentorship stay in Melbourne,” said Mr Perera.

“Our team hopes to transform our prototype into a commercial product for the market and believe that the mentorship program will provide us with the guidance and resources to turn our concept into reality.”

Team RegeX competed against a total of 70 teams from 18 universities in the Asia region across India, Sri Lanka, Indonesia, Pakistan, Vietnam and China. From those teams, 23 were successfully chosen as finalists and received up to AU$1000 to build their prototypes and receive ongoing professional feedback.

Other finalists also developed innovative prototypes such as: creating divergent wind turbines as a renewable energy source using existing infrastructures; and, designing smart glasses with real-time feedback of a person’s surroundings by detecting objects and audio input to improve accessibility and mobility.

This year the quality of entries was high, and the top teams had little that separated them, so it was decided to award cash prizes to the finalist teams. Team Drishti received AUD$2,000 for finishing 1st Runner up in the competition. The 2nd Runner up, Team Renewers, received AUD$1,000.  The People’s Choice award went to Team Mystic Technocrats who received AUD$2,000.

The Technology Infusion Grand Challenge competition is designed to encourage undergraduate students in Science, Technology, Engineering and Mathematics (STEM) and Business fields to be confident in their skills and knowledge to produce a solution and solve a concrete problem using next generation smart technologies.

La Trobe University extends their gratitude to the Victorian Government – Study Melbourne for being a co-sponsor of the Grand Challenge program.