UN MODELE CNN POUR LE CONTROLE D’ACCES DANS LES BATIMENTS INTELLIGENTS: CONCEPTION ET EVALUATION

In the context of increasing security demands for smart buildings, artificial intelligence particularly deep learning offers effective solutions for automated access management. This study proposes the design of a machine learning model based on a convolutional neural network (CNN) to develop an intelligent visual recognition access control system. The approach relies on analyzing images captured at access points to automatically and reliably identify authorized individuals. After presenting the fundamental theoretical concepts of machine learning and deep learning, the study details the implementation of a CNN model using the TensorFlow library. This includes dataset preparation, supervised training, and performance evaluation using relevant metrics. Experimental results demonstrate high accuracy in classifying normal and abnormal situations, confirming the approach’s potential for embedded security applications. Finally, this work highlights avenues for improvement, including multimodal data integration, real-time processing optimization, and enhanced personal data protection mechanisms.