ACADSTAFF UGM
| Title | : | Multiple Layered Deep Learning Based Real-time Face Detection |
| Author | : |
Muhammad Dwisnanto Putro (1) Wahyono, Ph.D. (2) Prof. Kanghyun Jo (3) |
| Date | : | 31 2019 |
| Keyword | : | Face detection,Real-time,Multiple layer,Deep learning,CPU Face detection,Real-time,Multiple layer,Deep learning,CPU |
| Abstract | : | This paper proposes an approach to real-time face detection working on the CPU. The challenge of the face detection system is the non-frontal face position and the use of accessories that cover the face area, even conventional detection systems that rely on facial features are difficult to get high accuracy. The proposed system can overcome these problems and work for multiple faces. The deep learning system can recognize facial features with complex backgrounds. The CNN (Convolution Neural Network) architecture with shallow layers to produce light computing then the system can work real-time. Multiple layer detection on the last feature map is used to detect varied face sizes. The system result shows sequential images of face localization. In the validation step, this system gets mAP is 90.84. In the testing process using the CPU, this system shows it can work real-time on various types of devices. |
| Group of Knowledge | : | |
| Level | : | Internasional |
| Status | : |
Published
|
| No | Title | Action |
|---|