Lightweight Model-Based Intrusion Detection in Construction Scenes

Abstract

Intrusion detection in construction scenes can effectively reduce the occurrence of hazardous incidents. Current detection methods, while effective, are often too complex. This paper proposes a lightweight monitoring model based on convolutional neural networks (CNNs). First, the model is trained using a dataset to achieve high accuracy. Then, the model is lightweighted using CNNs. Simulation results show that the model can maintain accuracy while occupying a smaller volume.