BROWSE
Detail
Conference Paper
딥러닝 기반 영상 정합 기술의 연구 동향
Cited - time in 
Share 
Share
- Issue Date
- 2021-11
- Citation
- 대한전자공학회 학술 대회 (추계) 2021, pp.454-457
- Publisher
- 대한전자공학회
- Language
- Korean
- Type
- Conference Paper
- Abstract
- In deep learning-based image stitching techniques, since the building datasets for training is very difficult, there have been many researchs on how to build datasets. The types of techniques can be classified according to the method of building a dataset, which can be largely divided into a fixed view technique and an arbitrary view technique. Because most of them are supervised learning models, photometric loss functions such as L1 or perceptual loss functions using pre-trained network are often used. In this paper, we review with a focus on how to build a dataset.
- KSP Keywords
- Learning-based, Supervised Learning, deep learning(DL), image stitching, learning models, loss function