VisDA: A Synthetic-to-Real Benchmark for Visual Domain Adaptation

Xingchao Peng, Ben Usman, Neela Kaushik, Dequan Wang, Judy Hoffman, Kate Saenko; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2018, pp. 2021-2026

We present the Synthetic-to-Real Visual Domain Adaptation (VisDA) Benchmark, a large-scale testbed for unsupervised domain adaptation across visual domains. The VisDA dataset is focused on the simulation-to-reality shift and has two associated tasks: image classification and image segmentation. The goal in both tracks is to first train a model on simulated, synthetic data in the source domain and then adapt it to perform well on real image data in the unlabeled test domain. Our dataset is the largest one to date for cross-domain object classification, with over 280K images across 12 categories in the combined training, validation and testing domains. The image segmentation dataset is also large-scale with over 30K images across 18 categories in the three domains. We compare VisDA to existing cross-domain adaptation datasets and provide a baseline performance analysis using various domain adaptation models that are currently popular in the field.