Luc Van Gool

TL;DR: An on-line learning scheme for Hough forests, which allows to extend their usage to further applications, such as the tracking of arbitrary target instances or large-scale learning of visual classifiers.

TL;DR: In this article, a cross-modal distillation framework is proposed for dense semantic labelling of sound-making objects, purely based on binaural sounds, which can be used for supervision transfer.

TL;DR: In this article, a deep learning method was proposed to detect and recognize keypoints in high-definition images at higher than real-time speed, and the detected two-dimensional keypoints along with their corresponding 3-D coordinates obtained from the robot sensors were used to calibrate the stereo microscope using an affine projection model.

TL;DR: The method, as a post-processing step, can significantly improve the quality of the hierarchical segmentation representations, and ease the usage of hierarchical image segmentation to high-level vision tasks such as object segmentation.

TL;DR: A novel method is proposed, which uses purely synthetic data to improve the performance on unseen real-world foggy scenes captured in the streets of Zurich and its surroundings, to highlight the potential and power of photo-realistic synthetic images for training and especially fine-tuning deep neural nets.