Delving Deeper into Anti-Aliasing in ConvNets
Xiao, Fanyi,Yu, Zhiding,Li, Yuheng
DOI: https://doi.org/10.1007/s11263-022-01672-y
IF: 13.369
2022-10-03
International Journal of Computer Vision
Abstract:Aliasing refers to the phenomenon that high frequency signals degenerate into completely different ones after sampling. It arises as a problem in the context of deep learning as downsampling layers are widely adopted in deep architectures to reduce parameters and computation. The standard solution is to apply a low-pass filter (e.g., Gaussian blur) before downsampling (Zhang in: ICML, 2020). However, it can be suboptimal to apply the same filter across the entire content, as the frequency of feature maps can vary across both spatial locations and feature channels. To tackle this, we propose an adaptive content-aware low-pass filtering layer, which predicts separate filter weights for each spatial location and channel group of the input feature maps. We investigate the effectiveness and generalization of the proposed method across multiple tasks, including image classification, semantic segmentation, instance segmentation, video instance segmentation, and image-to-image translation. Both qualitative and quantitative results demonstrate that our approach effectively adapts to the different feature frequencies to avoid aliasing while preserving useful information for recognition. Code is available at https://maureenzou.github.io/ddac/.
computer science, artificial intelligence