In the field of computer vision, human behaviour identification is a vital area of research with important applications in several fields, including as intelligent surveillance, smart homes, and virtual reality. The requirements for high recognition accuracy and applicability in the modern complex environment are difficult for traditional manual procedures to achieve. The advent of deep learning has created new opportunities for behaviour recognition research. The primary goal of this paper is behaviour recognition using convolutional neural networks (CNN). The study context and significance of behaviour recognition are initially discussed before delving into and examining the classical learning methods and deep learning methods of behaviour recognition. We build a convolution neural network-based algorithm based on the distinctive human behaviour in public spaces.