Sabine Süsstrunk | EPFL Graph Search

Prof. Dr. Sabine Süsstrunk leads the Image and Visual Representation Lab in the School of Computer and Communication Sciences (IC) at EPFL since 1999. From 2015-2020, she was also the first Director of the Digital Humanities Institute (DHI), College of Humanities (CdH). Her main research areas are in computational photography, computational imaging, color image processing and computer vision, machine learning, and computational image quality and aesthetics. Sabine has authored and co-authored over 200 publications, of which 7 have received best paper/demo awards, and holds over 10 patents. Sabine served as chair and/or committee member in many international conferences on image processing, computer vision, and image systems engineering. She is President of the Swiss Science Council SSC, Founding Member and Member of the Board (President 2014-2018) of the EPFL-WISH (Women in Science and Humanities) Foundation, Member of the Board of the SRG SSR (Swiss Radio and Television Corporation), and Member of the Board of Largo Films. She received the IS&T/SPIE 2013 Electronic Imaging Scientist of the Year Award for her contributions to color imaging, computational photography, and image quality, and the 2018 IS&T Raymond C. Bowman and the 2020 EPFL AGEPoly IC Polysphere Awards for excellence in teaching. Sabine is a Fellow of IEEE and IS&T.

DyNCA: Real-time Dynamic Texture Synthesis Using Neural Cellular Automata

Ehsan Pajouheshgar, Sabine Süsstrunk, Yitao Xu, Tong Zhang

Current Dynamic Texture Synthesis (DyTS) models can synthesize realistic videos. However, they require a slow iterative optimization process to synthesize a single fixed-size short video, and they do not offer any post-training control over the synthesis process. We propose Dynamic Neural Cellular Automata (DyNCA), a framework for real-time and controllable dynamic texture synthesis. Our method is built upon the recently introduced NCA models and can synthesize infinitely long and arbitrary-sized realistic video textures in real time. We quantitatively and qualitatively evaluate our model and show that our synthesized videos appear more realistic than the existing results. We improve the SOTA DyTS performance by

2\sim 4

orders of magnitude. Moreover, our model offers several real-time video controls including motion speed, motion direction, and an editing brush tool. We exhibit our trained models in an online interactive demo that runs on local hardware and is accessible on personal computers and smartphones. Link to the demo: https://dynca.github.io/

2023

TempSAL - Uncovering Temporal Information for Deep Saliency Prediction

Bahar Aydemir, Ludo Hoffstetter, Mathieu Salzmann, Sabine Süsstrunk, Tong Zhang

Deep saliency prediction algorithms complement the object recognition features, they typically rely on additional information, such as scene context, semantic relationships, gaze direction, and object dissimilarity. However, none of these models consider the temporal nature of gaze shifts during image observation. We introduce a novel saliency prediction model that learns to output saliency maps in sequential time intervals by exploiting human temporal attention patterns. Our approach locally modulates the saliency predictions by combining the learned temporal maps. Our experiments show that our method outperforms the state-of-the-art models, including a multi-duration saliency model, on the SALICON benchmark.

2023

Towards Stable and Efficient Adversarial Training against $l_1$ Bounded Adversarial Attacks

Yulun Jiang, Chen Liu, Mathieu Salzmann, Sabine Süsstrunk

We address the problem of stably and efficiently training a deep neural network robust to adversarial perturbations bounded by an

l_1

norm. We demonstrate that achieving robustness against

l_1

-bounded perturbations is more challenging than in the

l_2

l_\infty

cases, because adversarial training against

l_1

-bounded perturbations is more likely to suffer from catastrophic overfitting and yield training instabilities. Our analysis links these issues to the coordinate descent strategy used in existing methods. We address this by introducing Fast-EG-

l_1

, an efficient adversarial training algorithm based on Euclidean geometry and free of coordinate descent. Fast-EG-

l_1

comes with no additional memory costs and no extra hyper-parameters to tune. Our experimental results on various datasets demonstrate that Fast-EG-

l_1

yields the best and most stable robustness against

l_1

-bounded adversarial attacks among the methods of comparable computational complexity. Code and the checkpoints are available at https://github.com/IVRL/FastAdvL.

2023

DyNCA: Real-time Dynamic Texture Synthesis Using Neural Cellular Automata

Ehsan Pajouheshgar, Sabine Süsstrunk, Yitao Xu, Tong Zhang

2\sim 4

2023

Towards Stable and Efficient Adversarial Training against $l_1$ Bounded Adversarial Attacks

Yulun Jiang, Chen Liu, Mathieu Salzmann, Sabine Süsstrunk

We address the problem of stably and efficiently training a deep neural network robust to adversarial perturbations bounded by an

l_1

norm. We demonstrate that achieving robustness against

l_1

-bounded perturbations is more challenging than in the

l_2

l_\infty

cases, because adversarial training against

l_1

l_1

, an efficient adversarial training algorithm based on Euclidean geometry and free of coordinate descent. Fast-EG-

l_1

comes with no additional memory costs and no extra hyper-parameters to tune. Our experimental results on various datasets demonstrate that Fast-EG-

l_1

yields the best and most stable robustness against

l_1

-bounded adversarial attacks among the methods of comparable computational complexity. Code and the checkpoints are available at https://github.com/IVRL/FastAdvL.

2023