1 code implementation • 4 Mar 2024 • Puru Vaish, Shunxin Wang, Nicola Strisciuglio
However, common visual augmentations might not guarantee extensive robustness of computer vision models.
1 code implementation • 12 Aug 2023 • Shunxin Wang, Christoph Brune, Raymond Veldhuis, Nicola Strisciuglio
We propose a data augmentation strategy, named DFM-X, that leverages knowledge about frequency shortcuts, encoded in Dominant Frequencies Maps computed for image classification models.
1 code implementation • 28 Jul 2023 • Ioana Mazilu, Shunxin Wang, Sven Dummer, Raymond Veldhuis, Christoph Brune, Nicola Strisciuglio
We train autoencoders with implicit and explicit regularization techniques to enforce linearity relations among the representations of different blur levels in the latent space.
1 code implementation • ICCV 2023 • Shunxin Wang, Raymond Veldhuis, Christoph Brune, Nicola Strisciuglio
Our results demonstrate that NNs tend to find simple solutions for classification, and what they learn first during training depends on the most distinctive frequency characteristics, which can be either low- or high-frequencies.
1 code implementation • 10 May 2023 • Shunxin Wang, Raymond Veldhuis, Christoph Brune, Nicola Strisciuglio
The performance of computer vision models are susceptible to unexpected changes in input images, known as common corruptions (e. g. noise, blur, illumination changes, etc.