In this project, we develop ID card detection and recognition system which is called SeeU ID. SeeU ID is consist of OCR, image detection, face detection, and face recognition. The purpose of the task is to read the information from ID and verify the authority or accessibility for security.
To read the information, ID card must be detected. And ID card could not be laid in proper position. Therefore, we decide to detect ID card with segmentation model and after many review, we choose YOLOv11 segmentation model for this task.
To increase and ease of detecting and recognizing texts, we applied image alignment for image of ID card from segmentation model.
OCR model is used for this task. OCR model is consist of text detection and recognition. We tested and evaluated models and dbnet and svtr models are chosen.
(PS) During the model development, we tested many scheme such as changing the architecutre, modifing number of layers, activation function, loss function, and etc to improve the performance.