Rekayasa Perangkat Lunak Aplikasi Presensi Mobile Menggunakan Metode Deep Learning
Facial recognition research has its challenges due to faces complexity, ranging from facial expressions and certain conditions that make facial recognition an exciting research experiment. Moreover, many-oriented applications of machine learning have moved to devices edge, and-based facial recogniti...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Universitas Sebelas Maret
2023-12-01
|
| Series: | JIPTEK: Jurnal Ilmiah Pendidikan Teknik dan Kejuruan |
| Subjects: | |
| Online Access: | https://jurnal.uns.ac.id/jptk/article/view/76556 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Facial recognition research has its challenges due to faces complexity, ranging from facial expressions and certain conditions that make facial recognition an exciting research experiment. Moreover, many-oriented applications of machine learning have moved to devices edge, and-based facial recognition is no exception mobile. Seeing the ongoing development of facial pattern recognition algorithms such as Viola Jones, Backpropagation, this research uses the MobileFaceNet mobile CNN model which is currently popular to be implemented in the mobile-based facial recognition presence application at the Information and Computer Engineering Education (PTIK) FKIP UNS. The deep learning method is a method for understanding and classifying objects. In the developed application, a face is captured in an image. This research uses the help of the flutter framework and the Tensorflow Lite library to develop a presence application mobile facial recognition in real-time. This paper aims to determine the value of the memorization and generalization algorithms model of CNN MobileFaceNet on the application. A trial of the system has been carried out by involving 30 volunteers in the testing from 2016-2019 PTIK students by random sampling. Each test was carried out for 10 iterations. From the test results, the system memorization value is 84.5%. On the other hand, the generalization results get 70% in recognizing identical but not similar images correctly. In terms of memorization and generalization, these results are better than similar studies using backpropagation |
|---|---|
| ISSN: | 1979-0031 2598-6430 |