| Abstract |
Personal identification is the use of medical sciences and science knowledge to the identity of the person. The fingerprint is a method of identification that is convenient and effective. Since fingerprint verification wants analytical accuracy, artificial intelligence is used to increase the accuracy of the investigation. The population of Northeast Thailand has many ethnic groups. From 10 ethnic groups, classified into 2 speaking language families including Austro-Asian
language family or sub-family Mon-Khmer (mon, khmer, bon, so and suay) and Tai-Kadai language family (isan lao, phutai, kalereng, saek and yo). Therefore, fingerprint patterns were investigated of population and analysis of the relation. Which fingerprints were collected by rolled ink print technique from 406 people, a total of 4,060 fingerprints. Then scan fingerprints into image files and edit images from 1 person per 1 image to 1 person per 10 images (10 fingerprints). Then fingerprint analysis using AI. The dataset comprised 4,060 samples, of which 3,248 samples were used for training and 812 samples were used for testing. The results of training and testing accuracy of a convolutional neural network architecture across the various epochs. The proposed convolutional neural network classifier achieved 75.22% testing accuracy. |
Publication
Journal Publication
ไทย