For her poster on the paper "Distracting Downpour: Adversarial Weather Attacks for Motion Estimation", VIS researcher Jenny Schmalfuss received a best presentation prize (third place) at the International Computer Vision Summer School 2023 (ICVSS). The paper, which was co-authored with Lukas Mehl and Andrés Bruhn, presents a novel attack on motion estimation that exploits adversarially optimized particles to mimic weather effects like snowflakes, rain streaks, or fog clouds.
ICVSS 2023 took place in Sicily, Italy, from July 9-15, and accepted 170 out of 614 applicants. In total, three presentation prizes were awarded during the summer school. The paper on weather-based attacks on neural networks for motion estimation from image sequences, which was previewed at the summer school, will soon appear at one of the top conferences in computer vision: the International Conference on Computer Vision (ICCV), held from October 2-6, in Paris, France.