We extend single-view dataset to the multi-view edition as ‘DiLiGenT-MV’. This new dataset contains images of 5 objects of complex BRDFs taken from 20 views. and in each view, 96 calibrated point light sources are used. The ‘ground truth’ shape is available for quantitative evaluation. This ‘DiLiGenT-MV’ can be used to evaluate multi-view stereo methods under complex materials for lighting, be used to evaluate conventional single-view photometric stereo algorithms by treating each view independently.
Min Li, Zhenglong Zhou, Zhe Wu, Boxin Shi, Changyu Diao, Ping Tan, "Multi-View Photometric Stereo: A Robust Solution and Benchmark Dataset for Spatially Varying Isotropic Materials" , In IEEE Transactions on Image Processing (TIP), Volume 29, Issue 1, Pages 4159-4173, 2020. [PDF]
The ‘DiLiGenT-MV.zip’ package (6.85G) contains the following sub-folders:
‘mvpmsData’: ‘DiLiGenT-MV’ photometric stereo dataset with 5 objects (as shown in the banner). For each object, we provide 20 subfolders containing images from 20 viewpoints. The structure of every view subfolder is listed as follows: 16-bit integer PNG images with resolution of 612x512 from 96 different lighting directions; the mask image (‘mask.png’), lighting directions (‘light_directions.txt’, 3x96, with each row as a unit 3D vector), lighting intensities (‘light_intensities.txt’, 3x96, with each row represents the intensities in RGB channels; images are required to be normalized by dividing these intensity values per-channel before performing photometric stereo), and image file names (‘filenames.txt’); the ‘ground truth’ normals are stored as Matlab matrix (‘Normal_gt.mat’), and a color-encoded normal map is provided as a reference (‘Normal_gt.png’). The ‘ground truth’ meshes are stored as ply format (mesh_Gt.ply’). The calibrating information is stored as Matlab matrix (Calib_Results.mat’), which is acquired by using ‘Camera Calibration Toolbox for Matlab (Jean-Yves Bouguet )’.
‘estMesh’: Estimated the meshes stored as ply format by performing state-of-the-art multi-view photometric stereo methods on the ‘DiLiGenT-MV’ dataset; please refer to Section VI of the paper for details.
‘estNormal’: Estimated per-pixel surface normal values stored as Matlab matrices by performing state-of-the-art multi-view photometric stereo methods on the ‘DiLiGenT-MV’ dataset for each viewpoint; please refer to Section VI of the paper for details.
‘sampleCodeEvalMesh’: Sample codes that loads results in ‘estMesh’ and calculate shape error statistics (based on per-vertex distance error). The codes here reproduce Table Ⅲ of the paper.