The Korea Advanced Institute of Science and Technology (KAIST) announced that it has developed an artificial intelligence (AI) technology capable of detecting hidden illegal filming cameras using only a smartphone and a light-emitting diode (LED) device.
The SweepLED technology, developed by a research team led by Professor Han Jun from the School of Computing, analyzes the patterns of reflection changes appearing on the surface of an object by changing only the direction of the LED light while keeping the smartphone camera fixed.
While reflection points on general glossy objects change position or disappear depending on the direction of the lighting, camera lenses exhibit a characteristic where the reflection shape uniquely deforms due to the internal lens, aperture, and image sensor structure.
The research team trained a deep learning-based image analysis model on these time-dependent reflection change patterns to distinguish camera lenses.
While conventional detection methods relied on the user's eyes to simply search for "bright spots," SweepLED differentiates itself by having AI comprehensively analyze the movement and shape changes of reflections appearing from multiple lighting angles to identify hidden cameras.
Through this, it is possible to more stably detect camera lenses hidden inside various objects commonly found in accommodation spaces, such as chargers, desk clocks, remote controls, and ornaments.
When the research team evaluated the performance using 30 objects found in real-world environments, it recorded a detection accuracy of approximately 94%.
It was also confirmed that the time required to inspect a single object is under 5 seconds.
The research team explained that since the cost of the core components for the LED case attached to a smartphone is under 7 dollars (approx. 10,000 won), it could be commercialized in the future as an affordable detector in the form of a smartphone accessory or a product for general consumers.
Professor Han Jun said, "Illegal filming cameras are a serious threat to personal safety and privacy in everyday spaces," adding, "This is significant in that it presents the possibility of a practical detection technology that non-experts can easily utilize by combining smartphone-based low-cost hardware and artificial intelligence analysis."
(Photo courtesy of KAIST, Yonhap News)
※
Copying, redistribution, and unauthorized use in AI training are strictly prohibited.