Visual Material Traits: Trait Annotations
Kyoto University Computer Vision Lab
Visual material traits are locally-recognizable visual properties corresponding to characteristic material appearances. We have shown that these material traits can be used to identify materials based solely on locally-available image features.
We have augmented the Flickr Materials Database (FMD) of Sharan et al.1 with binary mask annotations for 13 visual material traits. We have used these annotations to learn visual material traits and, later, to evaluate automatically discovered local material attributes. We consistently annotated FMD images with material trait masks that highlight only local regions indiciative of each trait.
The traits we use in our dataset are:
Please cite the following work when you use this database in your research:
Visual Material Traits: Recognizing Per-Pixel Material Context
G. Schwartz and K. Nishino,
in Proc. of Color and Photometry in Computer Vision (Workshop held in conjunction with ICCV’13), Dec., 2013.
[ paper ][ database ][ project ]
Trait masks are stored in directories corresponding to the trait name. The filename of each mask corresponds to the original image in the FMD, simply replace the ‘.png’ extension with ‘.jpg’.
Download Material perception: What can you see in a brief glance?
L. Sharan, R. Rosenholtz, and E.H. Adelson,
Journal of Vision, vol. 14, no. 9, article 12, 2014.
[ Database ]