Learning from synthetic data for object re-identification can bridge the domain gap between synthetic and real-world data, enabling effective models.
