SweepLED helps users find hidden cameras with an attachable LED-array smartphone case. By sweeping illumination from different directions, it analyzes reflection changes and distinguishes camera lenses from ordinary objects.

Hidden cameras can be small enough to be concealed into useful everyday objects, yet this is also a reason why they can turn a hotel room, shared accommodation, restroom, or office into a serious privacy risk. To help users inspect suspicious objects more easily, a KAIST research team led by Professor Jun Han in the School of Computing developed SweepLED, a smartphone-based hidden camera detection system with an attachable LED-embedded case.

Many existing handheld detectors rely on a simple optical principle: shine light toward a suspicious object and look for a bright reflection from a camera lens. This method can be useful, but it often leaves the most difficult decision to the user. Everyday materials such as glass, metal, and glossy plastic can also produce bright spots. As a result, harmless objects such as chargers, clocks, remote controllers, or decorations may be mistaken for hidden cameras. 

SweepLED changes what the detector observes. Instead of moving the viewing position and the light source together, the system keeps the smartphone camera fixed and varies only the illumination direction. This creates decoupled illumination sweeps: light reaches the object from different directions while the viewing position remains the same. By separating the camera view from the lighting direction, SweepLED captures richer reflection changes from the same object.

The key insight is that ordinary shiny surfaces and camera lenses respond differently to changing light directions. A glossy surface usually creates a reflection that moves, fades, or disappears as the light direction changes. A camera lens, however, contains internal optical structures such as lens elements, an aperture, and an image sensor. These structures make the reflected pattern deform in a distinctive way.

SweepLED records these changes as a short video and analyzes them using an AI-based model. After the user scans a suspicious object, the system identifies regions that are likely to contain a hidden camera lens. This reduces the burden on users, who no longer need to judge ambiguous bright spots only by sight.

The research team evaluated SweepLED using 30 everyday objects that may appear in real accommodation environments. The system achieved about 94% detection accuracy and completed each object scan within a few seconds. The core components of the LED-embedded case cost less than seven U.S. dollars, suggesting that SweepLED could become a low-cost and accessible privacy protection tool.

SweepLED shows a practical direction for hidden camera detection in daily life. The study was led by KAIST in collaboration with the National University of Singapore and Singapore Management University, and was presented at ACM MobiSys 2026.