FIBO Sophomores Turn Standard CCTV into AI-Powered Accident Detectors

“I used to worry that AI would steal human jobs, so I decided to study AI myself that way, I won’t be replaced.” This simple yet witty logic from Ning explains the drive behind “CuddleCam: CCTV Security Project.” This AI system, designed to detect accidents, traffic violations, and solve congestion, earned the Teletubbies team the championship at the inaugural Muang Thong Hackathon 2024 in mid-2024. The project, developed by four second-year female students from the Institute of Field Robotics (FIBO) at King Mongkut’s University of Technology Thonburi (KMUTT), is now being supported for real-world implementation in Muang Thong Thani.

Chanyapak Supsawatkul (Ning), a Robotics and Automation Engineering student, shared that she and her teammates Natchanasa Lertmahakul (Eyejung), Bazla Siripattana (Buzz), and Nannaphat Nantapornnicha (Tick) chose this topic because of its direct benefit to the residents of Muang Thong Thani.

“The challenge set by Muang Thong Thani was to use AI to improve the quality of life for its residents. Before applying, we researched the community’s pain points online and found that traffic congestion and road accidents were major daily struggles. We wanted to empower existing CCTV cameras with AI to detect and alert authorities about accidents instantly,” said Ning.

The Teletubbies’ project was one of only 13 selected from 75 entries across 30 universities to reach the final round. During the competition, the students conducted deep field research, visiting key intersections to identify accident-prone spots and common violations like prohibited or double parking. They also observed the manual operations in the CCTV control room, confirming their belief that AI could significantly reduce the workload of staff and provide faster, more accurate traffic analysis—fitting perfectly into the Smart City vision.

The Innovation: High ROI and High Accuracy

Rather than just pitching an idea, the team presented a Minimum Viable Product (MVP) an AI prototype capable of analyzing car-to-car accidents from static images, complete with a user alert interface. Their Business Canvas analysis showed a high Return on Investment (ROI) because the solution focuses on software deployment, requiring minimal to no additional investment in existing CCTV hardware.

Following their victory, Muang Thong Thani signed a contract with the team to transition the MVP into a real-time AI system for field testing. Natchanasa (Eyejung) elaborated on the next steps: “We are training the AI with tens of thousands of image files to analyze real-time CCTV feeds for accidents and violations like crossing solid lines. We are also adding functions to count and classify vehicle types and evaluate traffic density, which will assist in long-term urban planning.”

Shifting Perspectives on AI

Dr. Rattanachai Ramathitima, a faculty member at FIBO and the project’s advisor, emphasized that beyond technical skills, it is crucial for students to have the right perspective on AI. He believes this project is a prime example of success driven by a deep understanding of user needs.

The project has transformed the students’ views as well. Ning moved from fearing AI to mastering it, while Buzz discovered that AI is a powerful tool for good. By reducing human error, they believe AI will foster a safer and more reliable society for everyone.