| Abstrakt: |
A person's information can be known through a personal document image. Personal document images are distinguished by variations in shape, size, color, and writing format. In an image captured by the camera, it is possible to have more than one type of personal document with variations in the object's tilt. In this case, the detection of objects for segmentation is followed by classification. This study proposes Convolutional Neural Networks (CNN) to classify four types of personal documents, namely Identity Card, Driving License, Vehicle Registration Number, and Passport. The experimental results show 98.18% accuracy with an experimental learning rate of 0.001. [ABSTRACT FROM AUTHOR] |
