Luc Van Gool

Bio: Luc Van Gool is an academic researcher from Katholieke Universiteit Leuven. The author has contributed to research in topics: Computer science & Object detection. The author has an hindex of 133, co-authored 1307 publications receiving 107743 citations. Previous affiliations of Luc Van Gool include Microsoft & ETH Zurich.

ETH-CVL @ MediaEval 2016: Textual-Visual Embeddings and Video2GIF for Video Interestingness.

Abstract: This paper presents the methods that underly our submission to the Predicting Media Interestingness Task at MediaEval 2016. Our contribution relies on two main approaches: (i) A similarity metric between image and text and (ii) a generic video highlight detector. In particular, we develop a method for learning the similarity of text and images, by projecting them into the same embedding space. This embedding allows to find video frames that are both, canonical and relevant w.r.t the title of the video. We present the result of different configurations and give insights into when our best performing method works well and where it has difficulties.

Modeling and synthesis of realistic visual speech in 3D

Abstract: The problem of realistic face animation is a difficult one. This is hampering a further breakthrough of some high-tech domains, such as special effects in the movies, the use of 3D face models in communications, the use of avatars and likenesses in virtual reality, and the production of games with more subtle scenarios. This work attempts to improve on the current stateof-the-art in face animation, especially for the creation of highly realistic lip and speech-related motions. To that end, 3D models of faces are used and based on the latest technology speech-related 3D face motion will be learned from examples. Thus, the chapter subscribes to the surging field of image-based modelling and widens its scope to include animation. The exploitation of detailed 3D motion sequences is quite unique, thereby Modeling and Synthesis of Realistic Visual Speech in 3D 267 Copyright © 2004, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc. is prohibited. narrowing the gap between modelling and animation. From measured 3D face deformations around the mouth area, typical motions are extracted for different “visemes.” Visemes are the basic motion patterns observed for speech and are comparable to the phonemes of auditory speech. The visemes are studied with sufficient detail to also cover natural variations and differences between individuals. Furthermore, the transition between visemes is analysed in terms of co-articulation effects, i.e., the visual blending of visemes as required for fluent, natural speech. The work presented in this chapter also encompasses the animation of faces for which no visemes have been observed and extracted. The “transplantation” of visemes to novel faces for which no viseme data have been recorded and for which only a static 3D model is available allows for the animation of faces without an extensive learning procedure for each individual.

Safe Motion Planning for Autonomous Driving using an Adversarial Road Model

Abstract: This paper presents a game-theoretic path-following formulation where the opponent is an adversary road model. This formulation allows us to compute safe sets using tools from viability theory, that can be used as terminal constraints in an optimization-based motion planner. Based on the adversary road model, we first derive an analytical discriminating domain, which even allows guaranteeing safety in the case when steering rate constraints are considered. Second, we compute the discriminating kernel and show that the output of the gridding based algorithm can be accurately approximated by a fully connected neural network, which can again be used as a terminal constraint. Finally, we show that by using our proposed safe sets, an optimization-based motion planner can successfully drive on city and country roads with prediction horizons too short for other baselines to complete the task.

Discovery of Sets of Mutually Orthogonal Vanishing Points in Videos

Abstract: While vanishing point (VP) estimation has received extensive attention, most approaches focus on static images or perform detection and tracking separately. In this paper, we focus on man-made environments and propose a novel method for detecting and tracking groups of mutually orthogonal vanishing points (MOVP), also known as Manhattan frames, jointly from monocular videos. The method is unique in that it is designed to enforce orthogonality in groups of VPs, temporal consistency of each individual MOVP, and orientation consistency of all putative MOVP. To this end, the method consists of three steps: 1) proposal of MOVP candidates by directly incorporating mutual orthogonality; 2) extracting consistent tracks of MOVPs by minimizing the flow cost over a network where nodes are putative MOVPs and edges are putative links across time; and 3) refinement of all MOVPs by enforcing consistency between lines, their identified vanishing directions and consistency of global camera orientation. The method is evaluated on six newly collected and annotated videos of urban scenes. Extensive experiments show that the method outperforms greedy MOVP tracking method considerably. In addition, we also test the method for camera orientation estimation and show that it obtains very promising results on a challenging street-view dataset.

A Novel Camera-Based System for Collaborative Interaction with Multi-Dimensional Data Models

Abstract: In this paper, we address the problem of effective visualization of and interaction with multiple and multi-dimensional data supporting communication between project stakeholders in an information cave. More exactly, our goal is to enable multiple users to interact with multiple screens from any location in an information cave. We present here our latest advancements in developing a novel human-computer interaction system that is specifically targeted towards room setups with physically spread sets of screens. Our system consists of a set of video cameras overseeing the room, and of which the signals are processed in real-time to detect and track the participants, their poses and hand-gestures. The system is fed with camera based gesture recognition. Early experiments have been conducted in the Value Lab (see figure 1), that has been recently introduced at ETH Zurich, and they focus on enabling the interaction with large urban 3D models being developed for the design and simulation of future cities. For the moment, experiments consider only the interaction of a single user with multiple layers (points of view) of a large city model displayed on multiple screens. The results demonstrate the huge potential of the system, and the principle of vision based interaction for such environments. The work continues on the extension of the system to a multi-user level.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Going deeper with convolutions

Abstract: We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.