image
imagewidth (px)
362
362
label
class label
23 classes
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick
0brick

Materials in Context Dataset (MINC-2500)

Dataset Summary

(from the website) MINC-2500 is a patch classification dataset with 2500 samples per category (Section 5.4 of the paper). This is a subset of MINC where samples have been sized to 362 x 362 and each category is sampled evenly. The original resolution images are not needed as we include the extracted patches in the archive.

Downloads last month
1
Edit dataset card