8 different 3D scenes have been rendered in Blender. Each scene is composed of two parts: Densely Sampled Light Field (DSLF) and ZScan. For each DSLF 51x51 images have been rendered, and 201 for each
Pre-computed sparse feature correspondences for pairs of images (outdoor and indoor) to reproduce the experiments described in our paper, particularly to train and evaluate...
8 different 3D scenes have been rendered in Blender. Each scene is composed of two parts: Densely Sampled Light Field (DSLF) and ZScan. For each DSLF 51x51 images have been rendered, and 201 for each
Extracted matches for evaluation of three view relative pose problems. This dataset was created for the paper "Practical Solutions to the Relative Pose of Three Calibrated Cameras" (CVPR 2025) To us