3 Computer Vision News Ruyu Wang optimized against producing defects, pose a challenge for collecting sufficient data, as waiting for defects to occur naturally could be a perpetual task. “The way I’m tackling this problem is by using a generative model to generate defective data for training – it’s synthetic data for data augmentation,” Ruyu tells us, having previously published another paper in this domain. “However, there is a gap between synthetic and real data. Models trained on synthetic data perform worse than those trained on real data.” This domain gap was first noticed in early works when synthetic data from GAN engines simulating scenes and products was used to train detection and BEST OF WACV 2024
RkJQdWJsaXNoZXIy NTc3NzU=