Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
Á¦·Î-¼¦ ÇнÀÀ» ÀÌ¿ëÇÑ ºñµð¿À ÇÁ·¹ÀÓ º¸°£¿¡ ´ëÇÑ Å½±¸ |
¿µ¹®Á¦¸ñ(English Title) |
Video Frame Interpolation Using Zero-shot Learning |
ÀúÀÚ(Author) |
°µ¿¿ì
ÀÌ»óÁø
ÀÌÇö¼º
ÀÌ»óÀ±
Dongwoo Kang
Sangjin Lee
Hean Sung Lee
Sangyoun Lee
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 45 NO. 01 PP. 2140 ~ 2142 (2022. 06) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Video frame interpolation is to synthesize the intermediate frame when consecutive frames have been provided. It requires an understanding of the complex and large motions in real-world video. AdaCoF is an algorithm of warping image for video frame interpolation. Zero-shot learning is a method of training a model with only one data rather than using prior knowledge learned with a dataset and inferring results on that data. In this paper, we propose a VFI model through zero-shot learning based on the AdaCoF algorithm. Although the performance of the model was not better than base model, it could be further developed based on this study.
|
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|