Integrated Deep Learning and Attention Mechanisms for Accurate License Plate Recognition

With the rapid increase in the number of motor vehicles, the related issues of vehicle management are also increasing, and license plate recognition has become one of the most important means of managing vehicles. However, traditional license plate recognition methods suffer from complex feature modeling and low recognition efficiency in complex scene conditions. Therefore, we propose a deep learning license plate recognition method based on improved ZFNet and a fused dual attention mechanism. First, the classic ZFNet is simplified to promote feature extraction efficiency. Then, the channel and spatial dual attention mechanisms are used to enrich the feature extraction for license plate character positions. Finally, the residual network module is utilized to improve license plate recognition accuracy. The experimental results demonstrate that the proposed approach achieves an average recognition accuracy of 98.92%, which is superior to other excellent deep learning models, such as EfficientNet and ResNet. The suggested strategy has high promotional significance for enhancing the artificial intelligence enabled automated vehicle scheduling and control in smart city transportation.