Semantic Segmentation on Road Images
Find the necessary data here: http://bair.berkeley.edu/blog/2018/05/30/bdd/
Rename all original images to "img (n)". Rename all color images to "seg (n)". Rename all label maps to "seg (n)".
n starts at 1.
There should be 7000 training images (/data/) and 1000 test images (/val/).