Performance evaluation of license plate detection using deep neural networks on NPU VIM3 hardware platform

The detection of license plates (LPs) is a crucial step to develop the intelligent traffic management systems. Several challenges exist for the detection of LPs such as the high variation of the geometry of LPs or the frequent variation in the conditions of LP image acquisition. The paper presents an end-to-end framework for the detection of LPs. The framework consists of two steps. The first one is the application and optimization of YOLOv4 network to detect LPs accurately. The second one is the strategy of the deployment and testing of the neural network on the NPU VIM3 tool kit. We have performed the evaluation on the large public dataset (Vietnamese license plate detection dataset). The performance comparison (the detection accuracy and execution time) with existed methods on various hardware platforms shows the effectiveness of the proposed method.