WiFire Dataset

Headlines · 1

1. Oct 18
   Swedish FOI researchers train WiFireNet to spot concealed handguns via Wi-Fi

   AI researchers at the Swedish Defense Research Agency (FOI) have developed WiFireNet, a system that identifies concealed handguns by analyzing channel state information from how objects distort ambient Wi-Fi signals, pitched as a non-intrusive alternative to visible cameras and metal detectors. The team built the WiFire Dataset by simultaneously recording RGB video and reflected Wi-Fi signals of people carrying objects, and trained a deep learning model on items including a plastic handgun, a metal water bottle, a real metal handgun and a phone. In controlled indoor trials tracking up to four moving people, the system averaged 72.5% handgun-detection precision, falling to 53.8% when required to pinpoint the weapon's location on the carrier, and it performs worse in unfamiliar layouts. The prototype used a modified 1999 IEEE 802.11a implementation, with Wi-Fi 7 and Wi-Fi 8 planned for later phases; the researchers open-sourced the code, models and dataset.

   Tom’s Hardware · 🔥 0

Experience and discussion from the community

Share my WiFire Dataset experienceAsk about WiFire Dataset