In a report by Cyber Insider, researchers have developed an AI system that can identify individuals carrying concealed handguns by analyzing Wi-Fi signals. This technology, named WiFireNet, aims to address limitations of current surveillance methods.
The WiFireNet system processes channel state information (CSI) from Wi-Fi signals, translating these measurements into a format that allows a neural network to predict the location of individuals and identify objects. Unlike cameras, which cannot see weapons hidden under clothing, or metal detectors that require passage through checkpoints and struggle to differentiate objects, Wi-Fi sensing offers a potential alternative. Experiments were conducted in controlled environments using a modified Wi-Fi setup. The system achieved an average precision of 72.5% for detecting handguns at a lenient threshold, decreasing to 53.8% with stricter localization requirements. While promising, the research acknowledges limitations, including performance in unfamiliar environments and potential privacy concerns due to its discreet screening capabilities. The researchers have released the dataset, model, and code to encourage further development in the field.
Source: Cyber Insider
