Transfer Learning in Hot Weapon Detection

Abstract

Due to the recent increase in dangerous crimes, the application of information technology in danger warning is very necessary, including smart cameras to help solve many problems of social context. This paper focuses on improving the performance of surveillance cameras to detect hot weapons i.e. particular knives and guns, in both accuracy and diversity by applying a transfer learning algorithm, which helps to provide early warning. in practical problems. The results of the paper show that the proposed method is positive with good accuracy when applied in practice, especially the model gives good accuracy with small data sets.