GTA-5

The 'GTA5' dataset is a benchmark that contains synthetic images and corresponding semantic labels used to evaluate the performance of unsupervised domain adaptation methods in semantic segmentation tasks.

Papers using GTA-5 (3)

Multiple Fusion Adaptation: A Strong Framework For Unsupervised Semantic Segmentation Adaptation2021 · 5 citesSSR: SAM Is A Strong Regularizer For Domain Adaptive Semantic Segmentation2024 · 2 citesSplit-merge Pooling2020