$7 Hidden Camera Detector Achieves 94% Accuracy with AI and LEDs

A research team led by the Korea Advanced Institute of Science and Technology (KAIST) has developed a system that combines a compact LED light that attaches to a smartphone with an AI app to detect hidden cameras with high accuracy. Priced at around $7 (about 10,000 won), it is attracting attention as a practical means of protecting privacy while traveling and in accommodations. As reported by Jowi Morales of Tom’s Hardware, the technology is described as overcoming the challenges of conventional manual detection methods.

The research was led by Professor Han Jun of the School of Computing at KAIST and was conducted as a joint study with the National University of Singapore (NUS) and Singapore Management University (SMU). The technology is named SweepLED, and related research documents have been made public. According to reporting by The Chosun Daily, its distinctive feature is that it consists of an LED grid that attaches magnetically to the back of a smartphone and an AI app that analyzes the captured images.

How SweepLED Works:

Detection Principle and Structure

At the core of SweepLED is an LED grid that is magnetically attached to the back of a smartphone. The grid continuously changes the position of the illuminated LEDs, sequentially switching the angle of illumination. Reflections that occur on various surfaces in a room move or disappear as the light source moves. In contrast, most cameras, including hidden cameras, have a distinctive structure consisting of a lens, aperture, and sensor, and therefore exhibit a distinctive pattern when reflecting light.

The research team focused on this difference. The AI app compares multiple images taken while changing the position of the lit LEDs and analyzes differences in how reflections move. While reflections from ordinary glossy surfaces change in tandem with the movement of the light source, reflections originating from a camera lens remain as a specific bright spot. The AI identifies this difference in behavior to determine whether a hidden camera is present.

As reported by Jowi Morales of Tom’s Hardware, “These lights change direction continuously, allowing the reflections on different surfaces to move or disappear altogether as they move around. Since most cameras, including hidden ones, have a unique structure with their lenses, aperture, and sensor, they often exhibit a distinctive reflection pattern which SweepLED can reliably detect.”

Conventional reflection-analysis apps rely on a single still image, and therefore often mistook glossy ornaments or metal parts for camera reflections. SweepLED is designed to isolate the causes of false detections by moving the light source itself and observing changes over time. This concept of dynamic illumination is what makes the combination of inexpensive hardware and AI possible.

Challenges and Limitations of Conventional

Detection Methods

Several methods for detecting hidden cameras using smartphones are already known. These include shining a flashlight around a room to look for tiny reflections from a lens, using a smartphone camera to capture infrared light emitted by hidden cameras, and using dedicated apps that analyze reflections.

However, the first two methods require users to manually scan the entire room themselves. Lenses are extremely small, only a few millimeters in size, and if cleverly hidden in gaps in furniture, ceilings, or inside power outlets, they are difficult to find visually. The possibility of overlooking them always remains, and the process itself is time-consuming.

While app-based reflection analysis is automated, concerns about its accuracy have been raised. Reflections under a single light source occur frequently even without a camera lens. Mirrors, glass, glossy plastics, and other reflective sources are abundant in accommodation interiors. A method that judges based only on a still image cannot distinguish these from hidden cameras, leading to frequent false positives. Users may become desensitized to warnings and risk overlooking real threats. The research team began developing SweepLED against the backdrop of the low practicality of these existing methods.

93.9% Accuracy Achieved in Demonstration

Experiments

To verify the effectiveness of SweepLED, the research team conducted evaluation experiments under conditions close to real-world environments. In the experiments, 30 items were prepared, with hidden cameras embedded in some of them, and detection was attempted. They recreated diverse installation patterns expected in daily life while varying the types of items, settings, and lighting conditions.

As a result, SweepLED achieved a detection accuracy of 93.9%. It is said to have shown significant improvements in both detection rate and reliability compared to manual scanning and conventional apps. The fact that it demonstrated practical accuracy despite being low-cost hardware at $7 is significant. Another advantage for widespread adoption is that it utilizes an existing general-purpose device — a smartphone — and does not require additional bulky equipment.

It should be noted that this experiment was conducted on a limited scale of 30 items and does not cover all environments or types of hidden cameras. Detection performance may still vary depending on lens size, concealment method, and ambient brightness. The research paper also suggests the need for further large-scale evaluation.

Background:

The Worsening Hidden Camera Problem in South Korea

The misuse of hidden cameras has emerged as a global privacy violation issue. Reports of harm have been particularly frequent in East Asia, and in South Korea it is widely recognized as a social problem. Cases of cameras being installed without permission in private spaces such as public restrooms, changing rooms, rental properties, and hotel rooms continue unabated.

In South Korea in 2018, large-scale protests led mainly by women called for stronger crackdowns on voyeuristic crimes and better protection for victims. Although measures have been taken, such as regular inspections of public facilities by the government and local authorities, the problem has not been eradicated as ultra-compact cameras have become readily available and concealment methods have become more sophisticated. For travelers as well, securing a means to verify the safety of their accommodations has become a pressing issue.

Against this backdrop, demand for detection tools that individuals can easily use is growing. Inspections by professional contractors are costly and time-consuming, making them difficult to use on a daily basis. There has been a need for a system that can inspect quickly and with high accuracy using an inexpensive accessory that can be retrofitted to a smartphone. SweepLED was designed to meet this demand.

Technical Limitations and Remaining Privacy

Challenges

While SweepLED has demonstrated high effectiveness in detecting hidden cameras, it does not solve all privacy protection challenges. Reports have highlighted issues such as more than 40,000 surveillance cameras worldwide streaming footage without adequate protection, and the spread of smart glasses that can record video and photos without those nearby noticing. These are areas that cannot be addressed by a method that detects lens reflections.

In addition, wearable devices with recording capabilities themselves are beginning to be seen as security concerns, as illustrated by reports that the U.S. Air Force has banned the use of certain devices across all units. Beyond detecting hidden cameras, comprehensive measures are needed that include consent from those being filmed and the channels through which data is distributed.

SweepLED’s detection depends solely on the optical reflection of a lens. It may be difficult to detect cameras that are powered off, cameras concealed by special methods that completely cover the lens, or voyeuristic methods that use lensless sensors. It is important to correctly understand the scope of the technology and avoid over-reliance.

Potential for Widespread Adoption Through Low Cost

The most distinctive feature of SweepLED is its extremely low manufacturing cost of $7. Its components are limited to an LED grid and a magnetic attachment mechanism, requiring no advanced sensors or dedicated processors. Because analysis is handled by the smartphone app and AI, it achieves high accuracy while keeping hardware complexity low.

This price range makes it possible to carry it for travel almost as a disposable item, or for accommodation operators to keep multiple units on hand for room inspections. Conventional commercially available hidden camera detectors cost from tens to hundreds of dollars and were not something ordinary travelers could purchase casually. If low-priced products like SweepLED become widespread, a new layer of defense — self-checks by users themselves — could be formed. As with the market expansion for compact hardware introduced in Punkt MC03 Shipping Begins, AYANEO Pocket Micro 2 Restocked, MINIX New AI Mini PC, peripherals for privacy protection are also expected to diversify in the future.

The research team has published a document detailing SweepLED, and further external verification and improvements are anticipated. While the timing of commercialization and specific sales format have not been disclosed, sharing the findings as open research could also accelerate the development of similar low-cost detection technologies.

Editorial Opinion

In the short term, we expect low-cost detectors like SweepLED to capture demand for self-inspections from travelers and the accommodation industry. The $7 price point and ease of use through smartphone integration could penetrate segments that conventional expensive dedicated devices could not reach. We also anticipate that accommodation booking platforms and travel-related services will begin to promote their adoption of such technologies as a safety measure.

In the long term, we believe there are limits to defense based on hardware alone as the cat-and-mouse game between detection and concealment continues. Threats that cannot be captured by reflection detection are growing, including lensless sensing, video leakage via networks, and unauthorized recording by wearable devices. We believe a multi-layered privacy protection framework will be required, including on-device recording detection, network monitoring, and the development of legal systems.

As a question from the editorial team, could the spread of personal detection tools反而 create a false sense of security? While 93.9% accuracy is high, in this domain the remaining few percent of missed detections can lead to serious harm. How should liability be handled when detection fails, and how should friction with facilities caused by false positives be addressed?

References

• ” Researchers build a $7 smartphone clip-on that spots hidden cameras — AI and dynamic LED grid deliver 94% accuracy ”, by Jowi Morales — Tom’s Hardware, 2026-09-02T12:20:00.000Z (ARR)
• Source URL: https://www.tomshardware.com/cameras/korean-researchers-build-usd7-hidden-camera-detector-gadget-uses-led-lights-and-ai-to-separate-reflections-from-lenses