Search Results for author: Thomas Brox

Found 96 papers, 45 papers with code

Multi-headed Neural Ensemble Search

no code implementations9 Jul 2021 Ashwin Raaghav Narayanan, Arber Zela, Tonmoy Saikia, Thomas Brox, Frank Hutter

Ensembles of CNN models trained with different seeds (also known as Deep Ensembles) are known to achieve superior performance over a single copy of the CNN.

Test-Time Adaptation to Distribution Shift by Confidence Maximization and Input Transformation

no code implementations28 Jun 2021 Chaithanya Kumar Mummadi, Robin Hutmacher, Kilian Rambach, Evgeny Levinkov, Thomas Brox, Jan Hendrik Metzen

This paper focuses on the fully test-time adaptation setting, where only unlabeled data from the target distribution is required.

Towards Total Recall in Industrial Anomaly Detection

2 code implementations15 Jun 2021 Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Schölkopf, Thomas Brox, Peter Gehler

Being able to spot defective parts is a critical component in large-scale industrial manufacturing.

 Ranked #1 on Anomaly Detection on MVTec AD (using extra training data)

Anomaly Detection Outlier Detection

Contrastive Representation Learning for Hand Shape Estimation

no code implementations8 Jun 2021 Christian Zimmermann, Max Argus, Thomas Brox

This work presents improvements in monocular hand shape estimation by building on top of recent advances in unsupervised learning.

Contrastive Learning Representation Learning

Pre-training of Deep RL Agents for Improved Learning under Domain Randomization

no code implementations29 Apr 2021 Artemij Amiranashvili, Max Argus, Lukas Hermann, Wolfram Burgard, Thomas Brox

Visual domain randomization in simulated environments is a widely used method to transfer policies trained in simulation to real robots.

Improving robustness against common corruptions with frequency biased models

no code implementations30 Mar 2021 Tonmoy Saikia, Cordelia Schmid, Thomas Brox

CNNs perform remarkably well when the training and test distributions are i. i. d, but unseen image corruptions can cause a surprisingly large drop in performance.

Data Augmentation Object Classification +1

What Causes Optical Flow Networks to be Vulnerable to Physical Adversarial Attacks

no code implementations30 Mar 2021 Simon Schrodi, Tonmoy Saikia, Thomas Brox

Recent work demonstrated the lack of robustness of optical flow networks to physical, patch-based adversarial attacks.

Optical Flow Estimation

On Exposing the Challenging Long Tail in Future Prediction of Traffic Actors

no code implementations23 Mar 2021 Osama Makansi, Özgün Cicek, Yassine Marrakchi, Thomas Brox

Predicting the states of dynamic traffic actors into the future is important for autonomous systems to operate safelyand efficiently.

Future prediction

COOT: Cooperative Hierarchical Transformer for Video-Text Representation Learning

1 code implementation NeurIPS 2020 Simon Ging, Mohammadreza Zolfaghari, Hamed Pirsiavash, Thomas Brox

Many real-world video-text tasks involve different levels of granularity, such as frames and words, clip and sentences or videos and paragraphs, each with distinct semantics.

Cross-Modal Retrieval Representation Learning +1

Beyond Single Stage Encoder-Decoder Networks: Deep Decoders for Semantic Image Segmentation

no code implementations19 Jul 2020 Gabriel L. Oliveira, Senthil Yogamani, Wolfram Burgard, Thomas Brox

In order to further improve the architecture we introduce a weight function which aims to re-balance classes to increase the attention of the networks to under-represented objects.

Optical Flow Estimation Semantic Segmentation

Scaling Imitation Learning in Minecraft

1 code implementation6 Jul 2020 Artemij Amiranashvili, Nicolai Dorka, Wolfram Burgard, Vladlen Koltun, Thomas Brox

Imitation learning is a powerful family of techniques for learning sensorimotor coordination in immersive environments.

Data Augmentation Imitation Learning +1

FlowControl: Optical Flow Based Visual Servoing

no code implementations1 Jul 2020 Max Argus, Lukas Hermann, Jon Long, Thomas Brox

One-shot imitation is the vision of robot programming from a single demonstration, rather than by tedious construction of computer code.

Optical Flow Estimation

Multimodal Future Localization and Emergence Prediction for Objects in Egocentric View with a Reachability Prior

1 code implementation CVPR 2020 Osama Makansi, Özgün Cicek, Kevin Buchicchio, Thomas Brox

In this paper, we investigate the problem of anticipating future dynamics, particularly the future location of other vehicles and pedestrians, in the view of a moving vehicle.

Temporal Shift GAN for Large Scale Video Generation

1 code implementation4 Apr 2020 Andres Munoz, Mohammadreza Zolfaghari, Max Argus, Thomas Brox

In this paper, we present a network architecture for video generation that models spatio-temporal consistency without resorting to costly 3D architectures.

Video Generation

Parting with Illusions about Deep Active Learning

no code implementations11 Dec 2019 Sudhanshu Mittal, Maxim Tatarchenko, Özgün Çiçek, Thomas Brox

Active learning aims to reduce the high labeling cost involved in training machine learning models on large datasets by efficiently labeling only the most informative samples.

Active Learning Data Augmentation +2

Adaptive Curriculum Generation from Demonstrations for Sim-to-Real Visuomotor Control

1 code implementation17 Oct 2019 Lukas Hermann, Max Argus, Andreas Eitel, Artemij Amiranashvili, Wolfram Burgard, Thomas Brox

We propose Adaptive Curriculum Generation from Demonstrations (ACGD) for reinforcement learning in the presence of sparse rewards.

Understanding and Robustifying Differentiable Architecture Search

1 code implementation ICLR 2020 Arber Zela, Thomas Elsken, Tonmoy Saikia, Yassine Marrakchi, Thomas Brox, Frank Hutter

Differentiable Architecture Search (DARTS) has attracted a lot of attention due to its simplicity and small search costs achieved by a continuous relaxation and an approximation of the resulting bi-level optimization problem.

Disparity Estimation Image Classification +1

Group Pruning using a Bounded-Lp norm for Group Gating and Regularization

no code implementations9 Aug 2019 Chaithanya Kumar Mummadi, Tim Genewein, Dan Zhang, Thomas Brox, Volker Fischer

We achieve state-of-the-art pruning results for ResNet-50 with higher accuracy on ImageNet.

AutoDispNet: Improving Disparity Estimation With AutoML

1 code implementation ICCV 2019 Tonmoy Saikia, Yassine Marrakchi, Arber Zela, Frank Hutter, Thomas Brox

In this work, we show how to use and extend existing AutoML techniques to efficiently optimize large-scale U-Net-like encoder-decoder architectures.

Disparity Estimation General Classification +1

Consistency-based anomaly detection with adaptive multiple-hypotheses predictions

2 code implementations ICLR 2019 Duc Tam Nguyen, Zhongyu Lou, Michael Klar, Thomas Brox

Thus, due to the lack of representative data, the wide-spread discriminative approaches cannot cover such learning tasks, and rather generative models, which attempt to learn the input density of the normal cases, are used.

Anomaly Detection

CAM-Convs: Camera-Aware Multi-Scale Convolutions for Single-View Depth

1 code implementation CVPR 2019 Jose M. Facil, Benjamin Ummenhofer, Huizhong Zhou, Luis Montesano, Thomas Brox, Javier Civera

Single-view depth estimation suffers from the problem that a network trained on images from one camera does not generalize to images taken with a different camera model.

3D Depth Estimation Domain Generalization

CrossNorm: Normalization for Off-Policy TD Reinforcement Learning

1 code implementation14 Feb 2019 Aditya Bhatt, Max Argus, Artemij Amiranashvili, Thomas Brox

Off-policy temporal difference (TD) methods are a powerful class of reinforcement learning (RL) algorithms.

Motion Perception in Reinforcement Learning with Dynamic Objects

no code implementations10 Jan 2019 Artemij Amiranashvili, Alexey Dosovitskiy, Vladlen Koltun, Thomas Brox

In dynamic environments, learned controllers are supposed to take motion into account when selecting the action to be taken.

Continuous Control

Defending Against Universal Perturbations With Shared Adversarial Training

no code implementations ICCV 2019 Chaithanya Kumar Mummadi, Thomas Brox, Jan Hendrik Metzen

Classifiers such as deep neural networks have been shown to be vulnerable against adversarial perturbations on problems with high-dimensional input space.

Image Classification Semantic Segmentation

Anomaly Detection With Multiple-Hypotheses Predictions

2 code implementations ICLR 2019 Duc Tam Nguyen, Zhongyu Lou, Michael Klar, Thomas Brox

In one-class-learning tasks, only the normal case (foreground) can be modeled with data, whereas the variation of all possible anomalies is too erratic to be described by samples.

Anomaly Detection

FusionNet and AugmentedFlowNet: Selective Proxy Ground Truth for Training on Unlabeled Images

no code implementations20 Aug 2018 Osama Makansi, Eddy Ilg, Thomas Brox

The latter can be used as proxy-ground-truth to train a network on real-world data and to adapt it to specific domains of interest.

Optical Flow Estimation

DeepTAM: Deep Tracking and Mapping

1 code implementation ECCV 2018 Huizhong Zhou, Benjamin Ummenhofer, Thomas Brox

For mapping, we accumulate information in a cost volume centered at the current depth estimate.

Depth Estimation

TD or not TD: Analyzing the Role of Temporal Differencing in Deep Reinforcement Learning

1 code implementation ICLR 2018 Artemij Amiranashvili, Alexey Dosovitskiy, Vladlen Koltun, Thomas Brox

Our understanding of reinforcement learning (RL) has been shaped by theoretical and empirical results that were obtained decades ago using tabular representations and linear function approximators.

ECO: Efficient Convolutional Network for Online Video Understanding

1 code implementation ECCV 2018 Mohammadreza Zolfaghari, Kamaljeet Singh, Thomas Brox

In this paper, we introduce a network architecture that takes long-term content into account and enables fast per-video processing at the same time.

Ranked #39 on Action Recognition on Something-Something V1 (using extra training data)

Action Classification Action Recognition +4

3D Human Pose Estimation in RGBD Images for Robotic Task Learning

no code implementations7 Mar 2018 Christian Zimmermann, Tim Welschehold, Christian Dornhege, Wolfram Burgard, Thomas Brox

We propose an approach to estimate 3D human pose in real world units from a single RGBD image and show that it exceeds performance of monocular 3D pose estimation approaches from color as well as pose estimation exclusively from depth.

3D Human Pose Estimation 3D Pose Estimation

Uncertainty Estimates and Multi-Hypotheses Networks for Optical Flow

1 code implementation ECCV 2018 Eddy Ilg, Özgün Çiçek, Silvio Galesso, Aaron Klein, Osama Makansi, Frank Hutter, Thomas Brox

Optical flow estimation can be formulated as an end-to-end supervised learning problem, which yields estimates with a superior accuracy-runtime tradeoff compared to alternative methodology.

Optical Flow Estimation

What Makes Good Synthetic Training Data for Learning Disparity and Optical Flow Estimation?

1 code implementation19 Jan 2018 Nikolaus Mayer, Eddy Ilg, Philipp Fischer, Caner Hazirbas, Daniel Cremers, Alexey Dosovitskiy, Thomas Brox

The finding that very large networks can be trained efficiently and reliably has led to a paradigm shift in computer vision from engineered solutions to learning formulations.

Optical Flow Estimation

Sparsity Invariant CNNs

1 code implementation22 Aug 2017 Jonas Uhrig, Nick Schneider, Lukas Schneider, Uwe Franke, Thomas Brox, Andreas Geiger

In this paper, we consider convolutional neural networks operating on sparse inputs with an application to depth upsampling from sparse laser scan data.

Depth Completion Depth Estimation

Artistic style transfer for videos and spherical images

4 code implementations13 Aug 2017 Manuel Ruder, Alexey Dosovitskiy, Thomas Brox

We propose a deep network architecture and training procedures that allow us to stylize arbitrary-length videos in a consistent and stable way, and nearly in real time.

Style Transfer Virtual Reality

End-to-End Learning of Video Super-Resolution with Motion Compensation

no code implementations3 Jul 2017 Osama Makansi, Eddy Ilg, Thomas Brox

We analyze the usage of optical flow for video super-resolution and find that common off-the-shelf image warping does not allow video super-resolution to benefit much from optical flow.

Motion Compensation Optical Flow Estimation +1

Topometric Localization with Deep Learning

no code implementations27 Jun 2017 Gabriel L. Oliveira, Noha Radwan, Wolfram Burgard, Thomas Brox

Compared to LiDAR-based localization methods, which provide high accuracy but rely on expensive sensors, visual localization approaches only require a camera and thus are more cost-effective while their accuracy and reliability typically is inferior to LiDAR-based methods.

Visual Localization Visual Odometry

Learning to Estimate 3D Hand Pose from Single RGB Images

8 code implementations ICCV 2017 Christian Zimmermann, Thomas Brox

Low-cost consumer depth cameras and deep learning have enabled reasonable 3D hand pose estimation from single depth images.

3D Hand Pose Estimation Sign Language Recognition

Universal Adversarial Perturbations Against Semantic Image Segmentation

no code implementations ICCV 2017 Jan Hendrik Metzen, Mummadi Chaithanya Kumar, Thomas Brox, Volker Fischer

We show empirically that there exist barely perceptible universal noise patterns which result in nearly the same predicted segmentation for arbitrary inputs.

Image Classification Semantic Segmentation

Octree Generating Networks: Efficient Convolutional Architectures for High-resolution 3D Outputs

1 code implementation ICCV 2017 Maxim Tatarchenko, Alexey Dosovitskiy, Thomas Brox

We present a deep convolutional decoder architecture that can generate volumetric 3D outputs in a compute- and memory-efficient manner by using an octree representation.

3D Reconstruction

Adversarial Examples for Semantic Image Segmentation

no code implementations3 Mar 2017 Volker Fischer, Mummadi Chaithanya Kumar, Jan Hendrik Metzen, Thomas Brox

Machine learning methods in general and Deep Neural Networks in particular have shown to be vulnerable to adversarial perturbations.

General Classification Image Classification +1

Hybrid Learning of Optical Flow and Next Frame Prediction to Boost Optical Flow in the Wild

no code implementations12 Dec 2016 Nima Sedaghat, Mohammadreza Zolfaghari, Thomas Brox

With the help of a sample-variant multi-tasking architecture, the network is trained on different tasks depending on the availability of ground-truth.

Action Classification Optical Flow Estimation

Protein contact prediction from amino acid co-evolution using convolutional networks for graph-valued images

no code implementations NeurIPS 2016 Vladimir Golkov, Marcin J. Skwark, Antonij Golkov, Alexey Dosovitskiy, Thomas Brox, Jens Meiler, Daniel Cremers

A contact map is a compact representation of the three-dimensional structure of a protein via the pairwise contacts between the amino acid constituting the protein.

Joint Graph Decomposition and Node Labeling: Problem, Algorithms, Applications

1 code implementation14 Nov 2016 Evgeny Levinkov, Jonas Uhrig, Siyu Tang, Mohamed Omran, Eldar Insafutdinov, Alexander Kirillov, Carsten Rother, Thomas Brox, Bernt Schiele, Bjoern Andres

In order to find feasible solutions efficiently, we define two local search algorithms that converge monotonously to a local optimum, offering a feasible solution at any time.

Combinatorial Optimization Multiple Object Tracking +2

Object Detection, Tracking, and Motion Segmentation for Object-level Video Segmentation

no code implementations10 Aug 2016 Benjamin Drayer, Thomas Brox

In contrast to most tracking methods, it provides an accurate, temporally consistent segmentation of each object.

Motion Segmentation Object Detection +4

A Multi-cut Formulation for Joint Segmentation and Tracking of Multiple Objects

no code implementations21 Jul 2016 Margret Keuper, Siyu Tang, Yu Zhongjie, Bjoern Andres, Thomas Brox, Bernt Schiele

Recently, Minimum Cost Multicut Formulations have been proposed and proven to be successful in both motion trajectory segmentation and multi-target tracking scenarios.

Motion Segmentation Object Detection

3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation

16 code implementations21 Jun 2016 Özgün Çiçek, Ahmed Abdulkadir, Soeren S. Lienkamp, Thomas Brox, Olaf Ronneberger

This paper introduces a network for volumetric segmentation that learns from sparsely annotated volumetric images.

Ranked #34 on 3D Part Segmentation on ShapeNet-Part (Instance Average IoU metric)

Data Augmentation

Point-wise mutual information-based video segmentation with high temporal consistency

no code implementations8 Jun 2016 Margret Keuper, Thomas Brox

In this paper, we tackle the problem of temporally consistent boundary detection and hierarchical segmentation in videos.

Boundary Detection Optical Flow Estimation +2

Synthesizing the preferred inputs for neurons in neural networks via deep generator networks

5 code implementations NeurIPS 2016 Anh Nguyen, Alexey Dosovitskiy, Jason Yosinski, Thomas Brox, Jeff Clune

Understanding the inner workings of such computational brains is both fascinating basic science that is interesting in its own right - similar to why we study the human brain - and will enable researchers to further improve DNNs.

Artistic style transfer for videos

4 code implementations28 Apr 2016 Manuel Ruder, Alexey Dosovitskiy, Thomas Brox

We present an approach that transfers the style from one image (for example, a painting) to a whole video sequence.

Style Transfer

Pixel-level Encoding and Depth Layering for Instance-level Semantic Labeling

no code implementations18 Apr 2016 Jonas Uhrig, Marius Cordts, Uwe Franke, Thomas Brox

Recent approaches for instance-aware semantic labeling have augmented convolutional neural networks (CNNs) with complex multi-task architectures or computationally expensive graphical models.

Instance Segmentation Semantic Segmentation

Video Segmentation With Just a Few Strokes

no code implementations ICCV 2015 Naveen Shankar Nagaraja, Frank R. Schmidt, Thomas Brox

As the use of videos is becoming more popular in computer vision, the need for annotated video datasets increases.

Motion Segmentation Video Segmentation +1

Unsupervised Generation of a Viewpoint Annotated Car Dataset From Videos

no code implementations ICCV 2015 Nima Sedaghat, Thomas Brox

Object recognition approaches have recently been extended to yield, aside of the object class output, also viewpoint or pose.

Object Recognition Structure from Motion

Global, Dense Multiscale Reconstruction for a Billion Points

no code implementations ICCV 2015 Benjamin Ummenhofer, Thomas Brox

We present a variational approach for surface reconstruction from a set of oriented points with scale information.

Multi-view 3D Models from Single Images with a Convolutional Network

no code implementations20 Nov 2015 Maxim Tatarchenko, Alexey Dosovitskiy, Thomas Brox

We present a convolutional network capable of inferring a 3D representation of a previously unseen object given a single image of this object.

Inverting Visual Representations with Convolutional Networks

2 code implementations CVPR 2016 Alexey Dosovitskiy, Thomas Brox

Inverting a deep network trained on ImageNet provides several insights into the properties of the feature representation learned by the network.

Learning to Generate Chairs With Convolutional Neural Networks

1 code implementation CVPR 2015 Alexey Dosovitskiy, Jost Tobias Springenberg, Thomas Brox

We train a generative convolutional neural network which is able to generate images of objects given object type, viewpoint, and color.

U-Net: Convolutional Networks for Biomedical Image Segmentation

374 code implementations18 May 2015 Olaf Ronneberger, Philipp Fischer, Thomas Brox

There is large consent that successful training of deep networks requires many thousand annotated training samples.

Cell Segmentation Colorectal Gland Segmentation: +7

Striving for Simplicity: The All Convolutional Net

33 code implementations21 Dec 2014 Jost Tobias Springenberg, Alexey Dosovitskiy, Thomas Brox, Martin Riedmiller

Most modern convolutional neural networks (CNNs) used for object recognition are built using the same principles: Alternating convolution and max-pooling layers followed by a small number of fully connected layers.

Image Classification Object Recognition

Learning to Generate Chairs, Tables and Cars with Convolutional Networks

1 code implementation21 Nov 2014 Alexey Dosovitskiy, Jost Tobias Springenberg, Maxim Tatarchenko, Thomas Brox

We train generative 'up-convolutional' neural networks which are able to generate images of objects given object style, viewpoint, and color.

Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks

1 code implementation26 Jun 2014 Alexey Dosovitskiy, Philipp Fischer, Jost Tobias Springenberg, Martin Riedmiller, Thomas Brox

While such generic features cannot compete with class specific features from supervised training on a classification task, we show that they are advantageous on geometric matching problems, where they also outperform the SIFT descriptor.

General Classification Geometric Matching

Spectral Graph Reduction for Efficient Image and Streaming Video Segmentation

no code implementations CVPR 2014 Fabio Galasso, Margret Keuper, Thomas Brox, Bernt Schiele

In contrast to previous work, the reduced graph is reweighted such that the resulting segmentation is equivalent, under certain assumptions, to that of the full graph.

Video Segmentation Video Semantic Segmentation

iPiano: Inertial Proximal Algorithm for Non-Convex Optimization

no code implementations18 Apr 2014 Peter Ochs, Yunjin Chen, Thomas Brox, Thomas Pock

A rigorous analysis of the algorithm for the proposed class of problems yields global convergence of the function values and the arguments.

Image Compression Image Denoising

Unsupervised feature learning by augmenting single images

no code implementations18 Dec 2013 Alexey Dosovitskiy, Jost Tobias Springenberg, Thomas Brox

We then extend these trivial one-element classes by applying a variety of transformations to the initial 'seed' patches.

Data Augmentation Object Recognition

An Iterated L1 Algorithm for Non-smooth Non-convex Optimization in Computer Vision

no code implementations CVPR 2013 Peter Ochs, Alexey Dosovitskiy, Thomas Brox, Thomas Pock

Here we extend the problem class to linearly constrained optimization of a Lipschitz continuous function, which is the sum of a convex function and a function being concave and increasing on the non-negative orthant (possibly non-convex and nonconcave on the whole space).

Image Denoising Optical Flow Estimation

Cannot find the paper you are looking for? You can Submit a new open access paper.