INTRODUCTION

Scene decomposition into its illuminant, shading, and reflectance intrinsic images is an essential step for scene understanding. Collecting intrinsic image groundtruth data is a laborious task. The assumptions on which the ground-truth procedures are based limit their application to simple scenes with a single object taken in the absence of indirect lighting and interreflections. We investigate synthetic data for intrinsic image research since extraction of ground truth is straightforward, and it allows for scenes in more realistic situations (e.g, with multiple illuminants and interreflections). With this dataset we aim to motivate researchers to further explore intrinsic image decomposition in complex scenes.

This dataset has been published in:
S. Beigpour, M. Serra, J. van de Weijer, R. Benavente, M. Vanrell, O.Penacchio, and D.Samaras. Intrinsic Image Evaluation on Synthetic Complex Scenes. IEEE International Conference on Image Processing (ICIP'2013), 2013. (pdf)

We evaluate the following three existing intrinsic image estimation algorithms on the data set:

1. M. Serra, O. Penacchio, R. Benavente, and M. Vanrell, Names and shades of color for intrinsic image estimation. , in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012. (webpage)

2. P.V. Gehler, C. Rother, M. Kiefel, L. Zhang, and B. Schölkopf, Recovering intrinsic images with a global sparsity prior on reflectance. , in Advances in Neural Information Processing Systems (NIPS), 2011. (webpage)

3. J.T. Barron and J. Malik, Color constancy, intrinsic images, and shape estimation. , in European Conference on Computer Vision (ECCV), 2012. (webpage)

Results estimation from these algorithms can be downloaded below.

DATA

Images of the dataset: synthetic_intrisic_image_dataset.rar

CODE

Matlab code to evaluate intrinsic image algorithms: code.rar

When combined with the 'Data' and the 'Results' files this code allows to generate the results table from the ICIP publication. Due to small implementation changes (different image rescaling) results differ slightly from those reported in ICIP publication (table.pdf ).

RESULT IMAGES

Results of three intrinsic image algorithms discussed above: results.rar