An object-centric version of Stylized COCO to benchmark texture bias and out-of-distribution robustness of vision models. See the ECCV 22 paper and supplementary material for details.
1 PAPER • NO BENCHMARKS YET