Random epipolar constraint loss functions for supervised optical flow estimation

Published: 2024, Last Modified: 13 May 2025Pattern Recognit. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•A loss function incorporates epipolar geometry into supervised optical flow models.•More interpretable and clear mathematical meaning for optical flow estimation.•A lightweight recurrent neural network for optical flow calculation.
Loading