A team of researchers from the Korea Advanced Institute of Science and Technology (KAIST), in collaboration with the National University of Singapore and Singapore Management University, has developed a low-cost smartphone accessory designed to detect hidden cameras. The device, named SweepLED, aims to address privacy concerns for individuals staying in hotels or vacation rentals.
SweepLED is a small LED accessory that clips onto a smartphone's camera. Its core components are estimated to cost less than $7 to build, making it a potentially accessible tool for the general public. The prototype utilizes 15 LEDs and connects to the back of a phone via a cable.
The system works by having the case-mounted LEDs sweep light onto a suspicious object from multiple angles while the smartphone records the reflections. The user points the phone at an object and holds it still at three different angles as the LEDs illuminate the area.
Unlike existing handheld camera detectors that often produce false positives from any glossy surface, SweepLED employs an AI-driven analysis. The software compares frames captured with the LEDs on against frames with the LEDs off to isolate reflections specifically caused by the sweep, filtering out ambient light and camera shake.
The system flags any reflection that appears lens-like. It then tracks how each candidate reflection behaves as the angle of the light shifts. The researchers explain that a curved or recessed shiny surface reflects light differently than a camera lens, and this difference, tracked frame by frame, is what a spatio-temporal AI model uses to make a determination. This analysis is cross-referenced across all three viewpoints captured by the user before a final decision is made.
Professor Jun Han of KAIST, who led the research, described the project as an effort to combine inexpensive smartphone hardware with AI analysis, enabling individuals without technical expertise to use it. The goal is to provide a practical tool for personal privacy without requiring specialized equipment or professional inspections.
Tests of SweepLED involved 30 objects: 12 items containing hidden cameras and 18 ordinary reflective items without cameras. The system achieved an overall accuracy of 93.9% in handheld settings and 95.1% in static settings. Researchers noted that its high accuracy in handheld scenarios underscores its practicality for real-world use.
The researchers concluded that SweepLED demonstrates potential as a practical and accessible tool for enhancing guest privacy in various accommodations, including hotels and short-term rentals. The system is designed to provide an answer regarding the presence of a hidden camera in under five seconds.
