Designing and Implementing a Real Time Weapons Detection System on ODROID-XU4

摘要

Artificial Intelligence has got many applications in security and camera analytics. In recent times, there are 245 million IP cameras already installed globally. To make these IP Cameras intelligent, processing power is required. For which we may use a cloud or a local hardware which has pros and cons. But in general, selecting a local hardware to run an AI application is more difficult and almost a challenging task. According to latest statistics gun ownerships among civilians on a rise. Latest statistics say 393 million guns are owned by civilians in USA, 71 million guns are owned by civilians in India and 50 million guns are owned by civilians in China etc., which says threats due to guns are increasing globally. Existing weapon detection systems are costly and requires special equipment like special scanners, cameras etc. Weapon detection systems that are simple and accurate are required. Hence, in this paper, we proposed a design and implementation of weapon detection system in a simple ODROID-XU4 connected to a cheap IP HD camera and achieved an accuracy more than 99.8% in real time. So as to give an alert to the nearest police men if any weapon found on busy roads or in heavy buildings.