Technology Developed to Detect Hidden Cameras Using Smartphone LEDs
Hidden cameras can be detected simply by attaching an LED to a smartphone. A research team led by Professor Han Jun from the School of Computing at KAIST announced on the 30th that they have developed 'SweepLED,' an artificial intelligence (AI) technology that identifies illegal filming cameras concealed in everyday items, through joint research with the National University of Singapore (NUS) and Singapore Management University (SMU).
With the smartphone camera pointed at a suspicious object, the LED sequentially changes the direction of the light it emits, while the continuous video captures the location, shape, and brightness changes of the reflection points. While ordinary glossy objects show moving or disappearing reflection points as the light direction changes, a camera lens exhibits unique reflection shapes and patterns depending on the light angle due to its internal optical structure, which consists of multiple layers of lenses, an aperture, and an image sensor. The research team first extracted reflection point candidates from the continuous video, analyzed their movement and morphological changes frame by frame, and used deep learning to distinguish between lenses and general reflective objects.
When a user points the smartphone at a suspicious object, a dedicated app controls both recording and analysis. Shake caused by holding the device by hand is corrected using video stabilization technology. If the smartphone is moved slightly to the side to film from multiple angles, the detection reliability can be further enhanced by combining analysis results from different viewpoints.
In experiments involving 30 different types of objects, including those containing cameras found indoors and general reflective objects, SweepLED achieved an overall average detection accuracy of approximately 94%, with an inspection time of about 4.5 seconds per object. High accuracy was confirmed in both nighttime and daytime indoor environments.
Since it analyzes the optical reflection of the lens rather than detecting wireless signals or electromagnetic waves, it can inspect cameras even if they are turned off or not recording or transmitting video, as long as the lens is exposed.
The research team produced a prototype in the form of a modular accessory that attaches to the back of a smartphone, similar to MagSafe. The device contains LEDs arranged in multiple directions and a small controller that sequentially controls the light. There is no need to modify the smartphone itself or connect separate detection equipment like a laptop. The manufacturing cost of core components, such as the LED array and printed circuit board (PCB), is less than $7, or around 10,000 won.
Professor Han Jun stated, "Illegal filming cameras pose a serious threat to personal safety and privacy in everyday spaces," adding, "We have demonstrated the potential of a detection technology that can be easily used by non-experts by combining low-cost smartphone devices with AI analysis."
KAIST doctoral student Yoon Jong-hyuk participated as the first author of this study, and the results were introduced at the international conference 'ACM MobiSys 2025' last June.
