Time-series Initialization and Conditioning for Video-agnostic Stabilization of Video Super-Resolution using Recurrent Networks

arXiv:2403.15832v1 Announce Type: new
Abstract: A Recurrent Neural Network (RNN) for Video Super Resolution (VSR) is generally trained with randomly clipped and cropped short videos extracted from original training videos due to various challenges in learning RNNs. However, since this RNN is optimized to super-resolve short videos, VSR of long videos is degraded due to the domain gap. Our preliminary experiments reveal that such degradation changes depending on the video properties, such as the video length and dynamics. To avoid …

abstract arxiv challenges cs.cv however network networks neural network recurrent neural network resolution rnn series super resolution training type video videos