Davis dataset for video object segmentation

The Davis dataset is a benchmark for video object segmentation. It contains 30 videos with 40 frames each.

BibTex: