Deblurred Image Quality Improvement in Learning-based Deblurring Datasets Utilizing Frame Interpolation
Masaki Aoi, Tomio Goto · 2023
The performance of machine learning in deblurring depends on the quality of the dataset used for training. In this paper, we propose a method to make the blurred images used for training closer to the real ones. Experimental results show that the model can achieve high performance in restoring blurred images if training is performed on blurred images created from videos of 480 fps or higher frame rates.