Personne
Sabine Süsstrunk
Publications associées (241)
Deep Gaussian denoiser epistemic uncertainty and decoupled dual-attention fusion
Sabine Süsstrunk, Majed El Helou, Xiaoyu Lin, Xiaoqi Ma
Following the performance breakthrough of denoising networks, improvements have come chiefly through novel architecture designs and increased depth. While novel denoising networks were designed for real images coming from different distributions, or for sp ...
2021Estimating Image Depth in the Comics Domain
Sabine Süsstrunk, Mathieu Salzmann, Martin Nicolas Everaert, Deblina Bhattacharjee
Estimating the depth of comics images is challenging as such images a) are monocular; b) lack ground-truth depth annotations; c) differ across different artistic styles; d) are sparse and noisy. We thus, use an off-the-shelf unsupervised image to image tra ...
2021Optimizing Latent Space Directions For GAN-based Local Image Editing
Sabine Süsstrunk, Tong Zhang, Ehsan Pajouheshgar
Generative Adversarial Network (GAN) based localized image editing can suffer from ambiguity between semantic attributes. We thus present a novel objective function to evaluate the locality of an image edit. By introducing the supervision from a pre-traine ...
IEEE2021Fidelity Estimation Improves Noisy-Image Classification with Pretrained Networks
Sabine Süsstrunk, Majed El Helou, Deblina Bhattacharjee, Xiaoyu Lin
Image classification has significantly improved using deep learning. This is mainly due to convolutional neural networks (CNNs) that are capable of learning rich feature extractors from large datasets. However, most deep learning classification methods are ...
2021NTIRE 2021 Depth Guided Image Relighting Challenge
Sabine Süsstrunk, Majed El Helou, Ruofan Zhou, Radu Timofte
Image relighting is attracting increasing interest due to its various applications. From a research perspective, image relighting can be exploited to conduct both image normalization for domain adaptation, and also for data augmentation. It also has multip ...
2021Adaptive extreme video completion
Sabine Süsstrunk, Radhakrishna Achanta, Majed El Helou, Ruofan Zhou
A computer-implemented video completion method is proposed for reconstructing a sparsely sampled video comprising sparsely sampled video frames comprising sparse sets of picture element intensity values. The method comprises: distributing the sparse sets o ...
2021On the Loss Landscape of Adversarial Training: Identifying Challenges and How to Overcome Them
Sabine Süsstrunk, Mathieu Salzmann, Tao Lin, Chen Liu
We analyze the influence of adversarial training on the loss landscape of machine learning models. To this end, we first provide analytical studies of the properties of adversarial loss functions under different adversarial budgets. We then demonstrate tha ...
2021Divergence-Based Adaptive Extreme Video Completion
Sabine Süsstrunk, Fabrice Jean Guibert, Majed El Helou, Ruofan Zhou, Frank Schmutz
Extreme image or video completion, where, for instance, we only retain 1% of pixels in random locations, allows for very cheap sampling in terms of the required pre-processing. The consequence is, however, a reconstruction that is challenging for humans an ...
2020Volumetric Transformer Networks
Sabine Süsstrunk, Mathieu Salzmann, Seungryong Kim
Existing techniques to encode spatial invariance within deep convolutional neural networks (CNNs) apply the same warping field to all the feature channels. This does not account for the fact that the individual feature channels can represent different sema ...
2020AIM 2020: Scene Relighting and Illumination Estimation Challenge
Sabine Süsstrunk, Xin Huang, Majed El Helou, Ruofan Zhou, Matthew Brown, Radu Timofte, Tongtong Zhao, Shanshan Zhao, Yu Zhu, Yuzhong Liu
We review the AIM 2020 challenge on virtual image relighting and illumination estimation. This paper presents the novel VIDIT dataset used in the challenge and the different proposed solutions and final evaluation results over the 3 challenge tracks. The f ...
2020