Traffic Sign Detection: Appropriate Data Augmentation Method from the Perspective of Frequency Domain

This study introduces a challenge faced by CNN in the task of traffic sign detection: how to achieve robustness distributional shift. At present, all kinds models rely on strong data augmentation methods enrich samples and robustness, such as Mosaic Mixup. In this study, we note that these do not have similar effects combating noise. We explore performance strategies against disturbance different frequency bands provide understanding from Fourier analysis perspective. can guidance for selecting detection tasks benchmark datasets.