The research group at the University of Amsterdam released a new challenging data set of synthetic garden images with ground-truth albedo, shading, and semantic annotations.

The data set is available at the website UVA Intrinsic Images and Semantic Segmentation Dataset, and from our data set page.

If you use the data set, please cite the following paper:
Anil S. Baslamisli, Thomas T. Groenestege, Partha Das, Hoang-An Le, Sezer Karaoglu, Theo Gevers, “Joint Learning of Intrinsic Images and Semantic Segmentation”, ECCV 2018: Computer Vision – ECCV 2018 pp 289-305