Abstract

Anomaly detection is the process of identifying irregular, unexpected, unpredictable, or unusual events or things that aren't considered to be typical occurrences of those events or typical occurrences of those things in a dataset and hence vary from established patterns. Security is always a top consideration in any sector due to an increase in crime rates in busy events or possibly lonely areas. In our study, automated gun/weapon identification is implemented using a convolution neural network. The suggested method uses two different kinds of datasets. The photographs in one dataset were labelled beforehand, whereas those in the second dataset required manual labelling. Although both algorithms achieve high levels of accuracy, their practical use may require a trade-off between speed and accuracy

Description