Identification of license plates is crucial in intelligent transportation systems, playing a key role in enhancing traffic control and security in various areas such as residential zones, corporate sectors, and government offices. This technology helps automate the monitoring of traffic systems and surveillance by recognizing and detecting vehicle number plates. The proposal aims to design a number plate recognition system based on convolutional neural network principles to improve security and surveillance. The proposal consists of four main steps: extraction of the license plate, preprocessing the image, segmenting the characters, and finally recognition of these characters. To improve character recognition, four advanced methods are explored: CNN, MobileNet, InceptionV3, and ResNet50, each with its own strengths and weaknesses. The proposal aims to identify the most effective technique based on the application and includes assessments of the model’s performance with real-time video feeds.