Smartphone add-on detected hidden camera lenses with 94% accuracy in tests

A smartphone-based system has identified hidden camera lenses with about 94% accuracy in tests on 30 everyday objects, taking no more than five seconds to inspect each one. Developed by researchers at KAIST, the National University of Singapore and Singapore Management University, SweepLED combines a phone camera with a small directional LED attachment and deep-learning analysis.

Instead of asking users to judge a single bright reflection, the system keeps the smartphone still while changing the direction of the LED light. Reflections from ordinary glossy materials tend to shift or disappear as the lighting angle changes. A camera lens produces a more distinctive sequence because of its internal lenses, aperture and image sensor. The AI was trained to recognize those changing shapes and movements, helping distinguish a concealed lens from metal, glass or shiny plastic.

The prototype found lenses hidden in objects such as chargers, desk clocks, remote controls and ornaments. The core components of its phone-mounted LED case cost less than US$7, although the sources do not report a commercial release. Presented at ACM MobiSys 2026, the work points toward an affordable consumer accessory that could give non-specialists a more objective way to inspect unfamiliar rooms and shared spaces.