Fast Data Generation for Training Deep-Learning 3D Reconstruction Approaches for Camera Arrays.

However, they require training datasets, of more or less significant size depending on the camera configuration, usually including ground truth information.

also in 312 other papers