Deep-Learning-Based Approach for Iraqi and Malaysian Vehicle License Plate Recognition

Figure 4

The architecture of the YOLOV2-based model and details of ResNet50 layers. The ResNet50 contains three main blocks, the input layers, the bottleneck layers, and residual (ResBlock) layers. For fine tuning, the pretrained deeper layers of ResNet50 are replaced with randomly initialized YOLO layers and trained again for LP segmentation and recognition.