| Title |
Lightweight Network for Mobile Real-time Frame Interpolation |
| Authors |
안현모(Hyeon Mo Ahn) ; 유광석(Kwang Seok Yoo) ; 이무재(Moo Jae Lee) ; 황원준(Wonjun Hwang) |
| DOI |
https://doi.org/10.5573/ieie.2023.60.5.27 |
| Keywords |
Deep learning; Video frame interpolation; Mobile; Optical flow; Real time |
| Abstract |
In this paper, we propose a neural network model and system for video frame interpolation that works in real time on mobile devices. With the development of technology, deep learning networks are being used in mobile devices in various fields as performance such as smartphones is advanced. However, as the performance of deep learning models increases, the required hardware resources are also increasing, making it difficult to import and use networks directly from mobile devices. For the implementation of a network running in a mobile device, we propose a network that minimizes performance loss and enables real-time processing through the distribution of lightweight techniques and operations. The CPU/GPU/NPU, which is the computing device of the smartphone, was tested for processing power and assigned to the computing device with optimal performance for each computing device so that it can be used in real time while maintaining network performance as much as possible. |