Several differences between the DTU datasets of MVSNet and PatchMatchNet must be noted

1 Test set data loading is different

1. The DTU test data set of MVSNet is different from the PatchmatchNet test data set;
the difference lies in data loading, the former cams file ends with the minimum depth and interval, and the latter is the minimum depth and maximum depth.

2 The resolution of the number of training sets is different

The training sets of the two are also different, the former is 19.2GB, and the latter is 76.3GB. The main difference is that the depth map of the former uses 119 groups of small resolution (160×128), and the depth map of the latter uses 128 groups of original resolution (1600×1200). .

DTU training set of MVSNet:
dtu_training.rar, 18.9GB;
mvs_training/dtu after decompression, 19.2GB;

MVSNet, Fsat-MVSNet, General

DTU training set for PatchMatchNet