Search Results for author: Daniel Cremers

Found 230 papers, 96 papers with code

Localizing Events in Videos with Multimodal Queries

no code implementations14 Jun 2024 Gengyuan Zhang, Mang Ling Ada Fok, Yan Xia, Yansong Tang, Daniel Cremers, Philip Torr, Volker Tresp, Jindong Gu

Our new benchmark aims to evaluate how well models can localize an event given a multimodal semantic query that consists of a reference image, which depicts the event, and a refinement text to adjust the images' semantics.

Video Understanding

An Image is Worth 32 Tokens for Reconstruction and Generation

no code implementations11 Jun 2024 Qihang Yu, Mark Weber, Xueqing Deng, Xiaohui Shen, Daniel Cremers, Liang-Chieh Chen

At ImageNet 512 x 512 benchmark, TiTok not only outperforms state-of-the-art diffusion model DiT-XL/2 (gFID 2. 74 vs. 3. 04), but also reduces the image tokens by 64x, leading to 410x faster generation process.

Power Variable Projection for Initialization-Free Large-Scale Bundle Adjustment

no code implementations8 May 2024 Simon Weber, Je Hyeong Hong, Daniel Cremers

This projective framework is crucial to solve large-scale bundle adjustment problem without initialization.

Uncertainty-Based Abstention in LLMs Improves Safety and Reduces Hallucinations

no code implementations16 Apr 2024 Christian Tomani, Kamalika Chaudhuri, Ivan Evtimov, Daniel Cremers, Mark Ibrahim

A major barrier towards the practical deployment of large language models (LLMs) is their lack of reliability.

Question Answering

Boosting Self-Supervision for Single-View Scene Completion via Knowledge Distillation

no code implementations CVPR 2024 Keonhee Han, Dominik Muhle, Felix Wimbauer, Daniel Cremers

In this work, we propose to fuse the scene reconstruction from multiple images and distill this knowledge into a more accurate single-view scene reconstruction.

Depth Estimation Depth Prediction +2

Finsler-Laplace-Beltrami Operators with Application to Shape Analysis

no code implementations CVPR 2024 Simon Weber, Thomas Dagès, Maolin Gao, Daniel Cremers

In experimental evaluations we demonstrate that the proposed FLBO is a valuable alternative to the traditional Riemannian-based LBO and ALBOs for spatial filtering and shape correspondence estimation.

Flattening the Parent Bias: Hierarchical Semantic Segmentation in the Poincaré Ball

1 code implementation4 Apr 2024 Simon Weber, Barış Zöngür, Nikita Araslanov, Daniel Cremers

Hierarchy is a natural representation of semantic taxonomies, including the ones routinely used in image segmentation.

Image Segmentation Segmentation +1

VXP: Voxel-Cross-Pixel Large-scale Image-LiDAR Place Recognition

no code implementations21 Mar 2024 Yun-Jin Li, Mariia Gladkova, Yan Xia, Rui Wang, Daniel Cremers

Recent works on the global place recognition treat the task as a retrieval problem, where an off-the-shelf global descriptor is commonly designed in image-based and LiDAR-based modalities.

Cross-Modal Retrieval Retrieval

Gaussian Splatting in Style

no code implementations13 Mar 2024 Abhishek Saroha, Mariia Gladkova, Cecilia Curreli, Tarun Yenamandra, Daniel Cremers

Scene stylization extends the work of neural style transfer to three spatial dimensions.

Style Transfer

Spectral Meets Spatial: Harmonising 3D Shape Matching and Interpolation

no code implementations CVPR 2024 Dongliang Cao, Marvin Eisenberger, Nafie El Amrani, Daniel Cremers, Florian Bernard

On the one hand, by incorporating spatial maps, our method obtains more accurate and smooth point-wise correspondences compared to previous functional map methods for shape matching.

Test-time Adaptation valid

Variational Learning is Effective for Large Deep Networks

1 code implementation27 Feb 2024 Yuesong Shen, Nico Daheim, Bai Cong, Peter Nickl, Gian Maria Marconi, Clement Bazan, Rio Yokota, Iryna Gurevych, Daniel Cremers, Mohammad Emtiyaz Khan, Thomas Möllenhoff

We give extensive empirical evidence against the common belief that variational learning is ineffective for large neural networks.

Sparse Views Near Light: A Practical Paradigm for Uncalibrated Point-light Photometric Stereo

no code implementations CVPR 2024 Mohammed Brahimi, Bjoern Haefner, Zhenzhang Ye, Bastian Goldluecke, Daniel Cremers

In contrast photometric stereo (PS) approaches have shown great potential for achieving high-quality reconstruction under sparse viewpoints.

3D Reconstruction

Deep Event Visual Odometry

1 code implementation15 Dec 2023 Simon Klenk, Marvin Motzet, Lukas Koestler, Daniel Cremers

To remove the dependency on additional sensors and to push the limits of using only a single event camera, we present Deep Event VO (DEVO), the first monocular event-only system with strong performance on a large number of real-world benchmarks.

Monocular Visual Odometry Pose Tracking

Erasing the Ephemeral: Joint Camera Refinement and Transient Object Removal for Street View Synthesis

no code implementations29 Nov 2023 Mreenav Shyam Deka, Lu Sang, Daniel Cremers

Synthesizing novel views for urban environments is crucial for tasks like autonomous driving and virtual tours.

Autonomous Driving

Text2Loc: 3D Point Cloud Localization from Natural Language

no code implementations CVPR 2024 Yan Xia, Letian Shi, Zifeng Ding, João F. Henriques, Daniel Cremers

We tackle the problem of 3D point cloud localization based on a few natural linguistic descriptions and introduce a novel neural network, Text2Loc, that fully interprets the semantic relationship between points and text.

Contrastive Learning

Enhancing Multimodal Compositional Reasoning of Visual Language Models with Generative Negative Mining

no code implementations7 Nov 2023 Ugur Sahin, Hang Li, Qadeer Khan, Daniel Cremers, Volker Tresp

Leveraging these generative hard negative samples, we significantly enhance VLMs' performance in tasks involving multimodal compositional reasoning.

HoloNets: Spectral Convolutions do extend to Directed Graphs

no code implementations3 Oct 2023 Christian Koke, Daniel Cremers

Within the graph learning community, conventional wisdom dictates that spectral convolutional networks may only be deployed on undirected graphs: Only there could the existence of a well-defined graph Fourier transform be guaranteed, so that information may be translated between spatial- and spectral domains.

Graph Learning Node Classification

ResolvNet: A Graph Convolutional Network with multi-scale Consistency

no code implementations30 Sep 2023 Christian Koke, Abhishek Saroha, Yuesong Shen, Marvin Eisenberger, Daniel Cremers

To remedy these shortcomings, we introduce ResolvNet, a flexible graph neural network based on the mathematical concept of resolvents.

Graph Learning Graph Neural Network

SCP: Scene Completion Pre-training for 3D Object Detection

no code implementations12 Sep 2023 Yiming Shan, Yan Xia, Yuhong Chen, Daniel Cremers

In this paper, we propose a Scene Completion Pre-training (SCP) method to enhance the performance of 3D object detectors with less labeled data.

3D Object Detection Autonomous Driving +2

Geometrically Consistent Partial Shape Matching

no code implementations10 Sep 2023 Viktoria Ehm, Paul Roetzer, Marvin Eisenberger, Maolin Gao, Florian Bernard, Daniel Cremers

Moreover, while in practice one often has only access to partial observations of a 3D shape (e. g. due to occlusion, or scanning artifacts), there do not exist any methods that directly address geometrically consistent partial shape matching.

Pose Transfer

Deep Video Codec Control for Vision Models

no code implementations30 Aug 2023 Christoph Reich, Biplob Debnath, Deep Patel, Tim Prangemeier, Daniel Cremers, Srimat Chakradhar

To overcome the deterioration of vision performance, this paper presents the first end-to-end learnable deep video codec control that considers both bandwidth constraints and downstream deep vision performance, while adhering to existing standardization.

Optical Flow Estimation Semantic Segmentation +1

SIGMA: Scale-Invariant Global Sparse Shape Matching

no code implementations ICCV 2023 Maolin Gao, Paul Roetzer, Marvin Eisenberger, Zorah Lähner, Michael Moeller, Daniel Cremers, Florian Bernard

We propose a novel mixed-integer programming (MIP) formulation for generating precise sparse correspondences for highly non-rigid shapes.

Robust Autonomous Vehicle Pursuit without Expert Steering Labels

no code implementations16 Aug 2023 Jiaxin Pan, Changyao Zhou, Mariia Gladkova, Qadeer Khan, Daniel Cremers

In this work, we present a learning method for lateral and longitudinal motion control of an ego-vehicle for vehicle pursuit.

Data Augmentation

Neural Poisson Surface Reconstruction: Resolution-Agnostic Shape Reconstruction from Point Clouds

1 code implementation3 Aug 2023 Hector Andrade-Loarca, Julius Hege, Daniel Cremers, Gitta Kutyniok

Overall, the neural Poisson surface reconstruction not only improves upon the limitations of classical deep neural networks in shape reconstruction but also achieves superior results in terms of reconstruction quality, running time, and resolution agnosticism.

3D Shape Reconstruction Super-Resolution +1

LiDAR View Synthesis for Robust Vehicle Navigation Without Expert Labels

1 code implementation2 Aug 2023 Jonathan Schmidt, Qadeer Khan, Daniel Cremers

We train a deep learning model, which takes a LiDAR scan as input and predicts the future trajectory as output.

Data Augmentation Self-Driving Cars

Learning Expressive Priors for Generalization and Uncertainty Estimation in Neural Networks

1 code implementation15 Jul 2023 Dominik Schnaus, JongSeok Lee, Daniel Cremers, Rudolph Triebel

In this work, we propose a novel prior learning method for advancing generalization and uncertainty estimation in deep neural networks.

Continual Learning Generalization Bounds

Enhancing Surface Neural Implicits with Curvature-Guided Sampling and Uncertainty-Augmented Representations

no code implementations3 Jun 2023 Lu Sang, Abhishek Saroha, Maolin Gao, Daniel Cremers

Neural implicits have become popular for representing surfaces because they offer an adaptive resolution and support arbitrary topologies.

Surface Reconstruction

Urban-StyleGAN: Learning to Generate and Manipulate Images of Urban Scenes

no code implementations16 May 2023 George Eskandar, Youssef Farag, Tarun Yenamandra, Daniel Cremers, Karim Guirguis, Bin Yang

Moreover, we employ an unsupervised latent exploration algorithm in the $\mathcal{S}$-space of the generator and show that it is more efficient than the conventional $\mathcal{W}^{+}$-space in controlling the image content.

Autonomous Driving Disentanglement +2

Learning Correspondence Uncertainty via Differentiable Nonlinear Least Squares

no code implementations CVPR 2023 Dominik Muhle, Lukas Koestler, Krishna Murthy Jatavallabhula, Daniel Cremers

We propose a differentiable nonlinear least squares framework to account for uncertainty in relative pose estimation from feature correspondences.

Pose Estimation

Scale-Equivariant Deep Learning for 3D Data

1 code implementation12 Apr 2023 Thomas Wimmer, Vladimir Golkov, Hoai Nam Dang, Moritz Zaiss, Andreas Maier, Daniel Cremers

The ability of convolutional neural networks (CNNs) to recognize objects regardless of their position in the image is due to the translation-equivariance of the convolutional operation.

Image Segmentation Medical Image Segmentation +1

Beyond In-Domain Scenarios: Robust Density-Aware Calibration

1 code implementation10 Feb 2023 Christian Tomani, Futa Waseda, Yuesong Shen, Daniel Cremers

While existing post-hoc calibration methods achieve impressive results on in-domain test datasets, they are limited by their inability to yield reliable uncertainty estimates in domain-shift and out-of-domain (OOD) scenarios.

Semidefinite Relaxations for Robust Multiview Triangulation

1 code implementation CVPR 2023 Linus Härenstam-Nielsen, Niclas Zeller, Daniel Cremers

We propose an approach based on convex relaxations for certifiably optimal robust multiview triangulation.

Behind the Scenes: Density Fields for Single View Reconstruction

1 code implementation CVPR 2023 Felix Wimbauer, Nan Yang, Christian Rupprecht, Daniel Cremers

By directly sampling color from the available views instead of storing color in the density field, our scene representation becomes significantly less complex compared to NeRFs, and a neural network can predict it in a single forward pass.

Depth Estimation Depth Prediction +1

Multi-Vehicle Trajectory Prediction at Intersections using State and Intention Information

1 code implementation6 Jan 2023 Dekai Zhu, Qadeer Khan, Daniel Cremers

This is done by training a neural network which takes the state and intent of the multiple vehicles to predict their future trajectory.

Trajectory Prediction

DDIT: Semantic Scene Completion via Deformable Deep Implicit Templates

no code implementations ICCV 2023 Haoang Li, Jinhu Dong, Binghui Wen, Ming Gao, Tianyu Huang, Yun-hui Liu, Daniel Cremers

It abstracts the shape prior of a category, and thus can provide constraints on the overall shape of an instance.

4Seasons: Benchmarking Visual SLAM and Long-Term Localization for Autonomous Driving in Challenging Conditions

no code implementations31 Dec 2022 Patrick Wenzel, Nan Yang, Rui Wang, Niclas Zeller, Daniel Cremers

In this paper, we present a novel visual SLAM and long-term localization benchmark for autonomous driving in challenging conditions based on the large-scale 4Seasons dataset.

Autonomous Driving Benchmarking +2

Masked Event Modeling: Self-Supervised Pretraining for Event Cameras

1 code implementation20 Dec 2022 Simon Klenk, David Bonello, Lukas Koestler, Nikita Araslanov, Daniel Cremers

The models pretrained with MEM are also label-efficient and generalize well to the dense task of semantic image segmentation.

 Ranked #1 on Classification on N-CARS (using extra training data)

Classification Event-based vision +2

SupeRVol: Super-Resolution Shape and Reflectance Estimation in Inverse Volume Rendering

no code implementations9 Dec 2022 Mohammed Brahimi, Bjoern Haefner, Tarun Yenamandra, Bastian Goldluecke, Daniel Cremers

We propose an end-to-end inverse rendering pipeline called SupeRVol that allows us to recover 3D shape and material parameters from a set of color images in a super-resolution manner.

Inverse Rendering Super-Resolution

PRISM: Probabilistic Real-Time Inference in Spatial World Models

no code implementations6 Dec 2022 Atanas Mirchev, Baris Kayalibay, Ahmed Agha, Patrick van der Smagt, Daniel Cremers, Justin Bayer

We introduce PRISM, a method for real-time filtering in a probabilistic generative model of agent motion and visual perception.

Bayesian Inference

G-MSM: Unsupervised Multi-Shape Matching with Graph-based Affinity Priors

no code implementations CVPR 2023 Marvin Eisenberger, Aysim Toker, Laura Leal-Taixé, Daniel Cremers

We present G-MSM (Graph-based Multi-Shape Matching), a novel unsupervised learning approach for non-rigid shape correspondence.

CASSPR: Cross Attention Single Scan Place Recognition

1 code implementation ICCV 2023 Yan Xia, Mariia Gladkova, Rui Wang, Qianyun Li, Uwe Stilla, João F. Henriques, Daniel Cremers

CASSPR uses queries from one branch to try to match structures in the other branch, ensuring that both extract self-contained descriptors of the point cloud (rather than one branch dominating), but using both to inform the output global descriptor of the point cloud.

A Graph Is More Than Its Nodes: Towards Structured Uncertainty-Aware Learning on Graphs

1 code implementation27 Oct 2022 Hans Hao-Hsun Hsu, Yuesong Shen, Daniel Cremers

Current graph neural networks (GNNs) that tackle node classification on graphs tend to only focus on nodewise scores and are solely evaluated by nodewise metrics.

Node Classification Structured Prediction

What Makes Graph Neural Networks Miscalibrated?

1 code implementation12 Oct 2022 Hans Hao-Hsun Hsu, Yuesong Shen, Christian Tomani, Daniel Cremers

Furthermore, based on the insights from this study, we design a novel calibration method named Graph Attention Temperature Scaling (GATS), which is tailored for calibrating graph neural networks.

Graph Attention Multi-class Classification

Deep Combinatorial Aggregation

1 code implementation12 Oct 2022 Yuesong Shen, Daniel Cremers

In this work, we explore a combinatorial generalization of deep ensemble called deep combinatorial aggregation (DCA).

Image Classification

E-NeRF: Neural Radiance Fields from a Moving Event Camera

1 code implementation24 Aug 2022 Simon Klenk, Lukas Koestler, Davide Scaramuzza, Daniel Cremers

We also show that combining events and frames can overcome failure cases of NeRF estimation in scenarios where only a few input views are available without requiring additional regularization.

Efficient and Flexible Sublabel-Accurate Energy Minimization

1 code implementation20 Jun 2022 Zhakshylyk Nurlanov, Daniel Cremers, Florian Bernard

While discrete optimization methods are able to give theoretical optimality guarantees, they can only handle a finite number of labels and therefore suffer from label discretization bias.

Image Denoising

Biologically Inspired Neural Path Finding

1 code implementation13 Jun 2022 Hang Li, Qadeer Khan, Volker Tresp, Daniel Cremers

The human brain can be considered to be a graphical structure comprising of tens of billions of biological neurons connected by synapses.

CHALLENGER: Training with Attribution Maps

no code implementations30 May 2022 Christian Tomani, Daniel Cremers

Regularization is key in deep learning, especially when training complex models on relatively small datasets.

Time Series Time Series Analysis

VPAIR -- Aerial Visual Place Recognition and Localization in Large-scale Outdoor Environments

1 code implementation23 May 2022 Michael Schleiss, Fahmi Rouatbi, Daniel Cremers

Visual Place Recognition and Visual Localization are essential components in navigation and mapping for autonomous vehicles especially in GNSS-denied navigation scenarios.

Self-Driving Cars Visual Localization +1

A Unified Framework for Implicit Sinkhorn Differentiation

1 code implementation CVPR 2022 Marvin Eisenberger, Aysim Toker, Laura Leal-Taixé, Florian Bernard, Daniel Cremers

The Sinkhorn operator has recently experienced a surge of popularity in computer vision and related fields.

Neural Implicit Representations for Physical Parameter Inference from a Single Video

no code implementations29 Apr 2022 Florian Hofherr, Lukas Koestler, Florian Bernard, Daniel Cremers

Neural networks have recently been used to analyze diverse physical systems and to identify the underlying dynamics.

Power Bundle Adjustment for Large-Scale 3D Reconstruction

2 code implementations CVPR 2023 Simon Weber, Nikolaus Demmel, Tin Chon Chan, Daniel Cremers

We demonstrate that employing the proposed Power Bundle Adjustment as a sub-problem solver significantly improves speed and accuracy of the distributed optimization.

3D Reconstruction Distributed Optimization

A Scalable Combinatorial Solver for Elastic Geometrically Consistent 3D Shape Matching

1 code implementation CVPR 2022 Paul Roetzer, Paul Swoboda, Daniel Cremers, Florian Bernard

We present a scalable combinatorial algorithm for globally optimizing over the space of geometrically consistent mappings between 3D shapes.

Implicit Shape Completion via Adversarial Shape Priors

no code implementations21 Apr 2022 Abhishek Saroha, Marvin Eisenberger, Tarun Yenamandra, Daniel Cremers

Finally, we show that our adversarial training approach leads to visually plausible reconstructions that are highly consistent in recovering missing parts of a given object.

Point Cloud Completion

The Probabilistic Normal Epipolar Constraint for Frame-To-Frame Rotation Optimization under Uncertain Feature Positions

no code implementations CVPR 2022 Dominik Muhle, Lukas Koestler, Nikolaus Demmel, Florian Bernard, Daniel Cremers

However, their approach does not take into account uncertainties, so that the accuracy of the estimated relative pose is highly dependent on accurate feature positions in the target frame.

Lateral Ego-Vehicle Control without Supervision using Point Clouds

no code implementations20 Mar 2022 Florian Müller, Qadeer Khan, Daniel Cremers

In this paper, a framework for training a more robust and scalable model for lateral vehicle control is proposed.

Visual Odometry

Intrinsic Neural Fields: Learning Functions on Manifolds

1 code implementation15 Mar 2022 Lukas Koestler, Daniel Grittner, Michael Moeller, Daniel Cremers, Zorah Lähner

Neural fields have gained significant attention in the computer vision community due to their excellent performance in novel view synthesis, geometry reconstruction, and generative modeling.

Novel View Synthesis

Vision-based Large-scale 3D Semantic Mapping for Autonomous Driving Applications

no code implementations2 Mar 2022 Qing Cheng, Niclas Zeller, Daniel Cremers

In this paper, we present a complete pipeline for 3D semantic mapping solely based on a stereo camera system.

Autonomous Driving Visual Odometry

DM-VIO: Delayed Marginalization Visual-Inertial Odometry

1 code implementation11 Jan 2022 Lukas von Stumberg, Daniel Cremers

This is the foundation of the proposed pose graph bundle adjustment, which we use for IMU initialization.

Shortest Paths in Graphs with Matrix-Valued Edges: Concepts, Algorithm and Application to 3D Multi-Shape Analysis

1 code implementation8 Dec 2021 Viktoria Ehm, Daniel Cremers, Florian Bernard

Traditionally, the concept of a shortest path is considered for graphs with scalar edge weights, which makes it possible to compute the length of a path by adding up the individual edge weights.

Image Segmentation Semantic Segmentation

Gradient-SDF: A Semi-Implicit Surface Representation for 3D Reconstruction

1 code implementation CVPR 2022 Christiane Sommer, Lu Sang, David Schubert, Daniel Cremers

We present Gradient-SDF, a novel representation for 3D geometry that combines the advantages of implict and explicit representations.

3D Reconstruction

Multidirectional Conjugate Gradients for Scalable Bundle Adjustment

no code implementations8 Oct 2021 Simon Weber, Nikolaus Demmel, Daniel Cremers

We revisit the problem of large-scale bundle adjustment and propose a technique called Multidirectional Conjugate Gradients that accelerates the solution of the normal equation by up to 61%.

Sparse Quadratic Optimisation over the Stiefel Manifold with Application to Permutation Synchronisation

no code implementations NeurIPS 2021 Florian Bernard, Daniel Cremers, Johan Thunberg

We address the non-convex optimisation problem of finding a sparse matrix on the Stiefel manifold (matrices with mutually orthogonal columns of unit length) that maximises (or minimises) a quadratic objective function.

Constrained Clustering Graph Matching

Scalable Sinkhorn Backpropagation

no code implementations29 Sep 2021 Marvin Eisenberger, Aysim Toker, Laura Leal-Taixé, Florian Bernard, Daniel Cremers

Our main contribution is deriving a simple and efficient algorithm that performs this backward pass in closed form.

Rolling Shutter Correction

Scene Graph Generation for Better Image Captioning?

no code implementations23 Sep 2021 Maximilian Mozes, Martin Schmitt, Vladimir Golkov, Hinrich Schütze, Daniel Cremers

We investigate the incorporation of visual relationships into the task of supervised image caption generation by proposing a model that leverages detected objects and auto-generated visual relationships to describe images in natural language.

Caption Generation Graph Generation +2

Towards Robust Monocular Visual Odometry for Flying Robots on Planetary Missions

1 code implementation12 Sep 2021 Martin Wudenka, Marcus G. Müller, Nikolaus Demmel, Armin Wedler, Rudolph Triebel, Daniel Cremers, Wolfgang Stürzl

In contrast to most other approaches, our framework can also handle rotation-only motions that are particularly challenging for monocular odometry systems.

Ingenuity Monocular Visual Odometry +1

Square Root Marginalization for Sliding-Window Bundle Adjustment

no code implementations ICCV 2021 Nikolaus Demmel, David Schubert, Christiane Sommer, Daniel Cremers, Vladyslav Usenko

The square root formulation pervades three major aspects of our optimization-based sliding-window estimator: for bundle adjustment we eliminate landmark variables with nullspace projection; to store the marginalization prior we employ a matrix square root of the Hessian; and when marginalizing old poses we avoid forming normal equations and update the square root prior directly with a specialized QR decomposition.

Dive into Layers: Neural Network Capacity Bounding using Algebraic Geometry

1 code implementation3 Sep 2021 Ji Yang, Lu Sang, Daniel Cremers

To mathematically prove this, we borrow a tool in topological algebra: Betti numbers to measure the topological geometric complexity of input data and the neural network.

TUM-VIE: The TUM Stereo Visual-Inertial Event Dataset

no code implementations16 Aug 2021 Simon Klenk, Jason Chui, Nikolaus Demmel, Daniel Cremers

The event cameras contain a large sensor of 1280x720 pixels, which is significantly larger than the sensors used in existing stereo event datasets (at least by a factor of ten).

Explicit Pairwise Factorized Graph Neural Network for Semi-Supervised Node Classification

no code implementations27 Jul 2021 Yu Wang, Yuesong Shen, Daniel Cremers

To learn the direct influence among output nodes in a graph, we propose the Explicit Pairwise Factorized Graph Neural Network (EPFGNN), which models the whole graph as a partially observed Markov Random Field.

Graph Neural Network Node Classification

Lifting the Convex Conjugate in Lagrangian Relaxations: A Tractable Approach for Continuous Markov Random Fields

no code implementations13 Jul 2021 Hartmut Bauermeister, Emanuel Laude, Thomas Möllenhoff, Michael Moeller, Daniel Cremers

In contrast to existing discretizations which suffer from a grid bias, we show that a piecewise polynomial discretization better preserves the continuous nature of our problem.

Stereo Matching

Event-Based Feature Tracking in Continuous Time with Sliding Window Optimization

no code implementations9 Jul 2021 Jason Chui, Simon Klenk, Daniel Cremers

We propose a novel method for continuous-time feature tracking in event cameras.

NeuroMorph: Unsupervised Shape Interpolation and Correspondence in One Go

no code implementations CVPR 2021 Marvin Eisenberger, David Novotny, Gael Kerchenbaum, Patrick Labatut, Natalia Neverova, Daniel Cremers, Andrea Vedaldi

We present NeuroMorph, a new neural network architecture that takes as input two 3D shapes and produces in one go, i. e. in a single feed forward pass, a smooth interpolation and point-to-point correspondences between them.

DeepLab2: A TensorFlow Library for Deep Labeling

4 code implementations17 Jun 2021 Mark Weber, Huiyu Wang, Siyuan Qiao, Jun Xie, Maxwell D. Collins, Yukun Zhu, Liangzhe Yuan, Dahun Kim, Qihang Yu, Daniel Cremers, Laura Leal-Taixe, Alan L. Yuille, Florian Schroff, Hartwig Adam, Liang-Chieh Chen

DeepLab2 is a TensorFlow library for deep labeling, aiming to provide a state-of-the-art and easy-to-use TensorFlow codebase for general dense pixel prediction problems in computer vision.

Joint Deep Multi-Graph Matching and 3D Geometry Learning from Inhomogeneous 2D Image Collections

no code implementations31 Mar 2021 Zhenzhang Ye, Tarun Yenamandra, Florian Bernard, Daniel Cremers

While these approaches mainly focus on learning node and edge attributes, they completely ignore the 3D geometry of the underlying 3D objects depicted in the 2D images.

Graph Matching

Self-Supervised Steering Angle Prediction for Vehicle Control Using Visual Odometry

no code implementations20 Mar 2021 Qadeer Khan, Patrick Wenzel, Daniel Cremers

Vision-based learning methods for self-driving cars have primarily used supervised approaches that require a large number of labels for training.

Self-Driving Cars Visual Odometry

Square Root Bundle Adjustment for Large-Scale Reconstruction

1 code implementation CVPR 2021 Nikolaus Demmel, Christiane Sommer, Daniel Cremers, Vladyslav Usenko

We propose a new formulation for the bundle adjustment problem which relies on nullspace marginalization of landmark variables by QR decomposition.

Parameterized Temperature Scaling for Boosting the Expressive Power in Post-Hoc Uncertainty Calibration

1 code implementation24 Feb 2021 Christian Tomani, Daniel Cremers, Florian Buettner

We address the problem of uncertainty calibration and introduce a novel calibration method, Parametrized Temperature Scaling (PTS).

Variational Data Assimilation with a Learned Inverse Observation Operator

1 code implementation22 Feb 2021 Thomas Frerix, Dmitrii Kochkov, Jamie A. Smith, Daniel Cremers, Michael P. Brenner, Stephan Hoyer

Variational data assimilation optimizes for an initial state of a dynamical system such that its evolution fits observational data.

Weather Forecasting

Tight Integration of Feature-based Relocalization in Monocular Direct Visual Odometry

no code implementations1 Feb 2021 Mariia Gladkova, Rui Wang, Niclas Zeller, Daniel Cremers

To achieve map-based relocalization for direct methods, we integrate image features into Direct Sparse Odometry (DSO) and rely on feature matching to associate online visual odometry (VO) with a previously built map.

Visual Odometry

Post-hoc Uncertainty Calibration for Domain Drift Scenarios

1 code implementation CVPR 2021 Christian Tomani, Sebastian Gruber, Muhammed Ebrar Erdem, Daniel Cremers, Florian Buettner

First, we show that existing post-hoc calibration methods yield highly over-confident predictions under domain shift.

Neural Online Graph Exploration

1 code implementation6 Dec 2020 Ioannis Chiotellis, Daniel Cremers

Can we learn how to explore unknown spaces efficiently?

Future prediction

Isometric Multi-Shape Matching

no code implementations CVPR 2021 Maolin Gao, Zorah Lähner, Johan Thunberg, Daniel Cremers, Florian Bernard

Finding correspondences between shapes is a fundamental problem in computer vision and graphics, which is relevant for many applications, including 3D reconstruction, object tracking, and style transfer.

3D Reconstruction Object Tracking +1

i3DMM: Deep Implicit 3D Morphable Model of Human Heads

1 code implementation CVPR 2021 Tarun Yenamandra, Ayush Tewari, Florian Bernard, Hans-Peter Seidel, Mohamed Elgharib, Daniel Cremers, Christian Theobalt

Our approach has the following favorable properties: (i) It is the first full head morphable model that includes hair.

Non-Rigid Puzzles

no code implementations26 Nov 2020 Or Litany, Emanuele Rodolà, Alex Bronstein, Michael Bronstein, Daniel Cremers

We assume to be given a reference shape and its multiple parts undergoing a non-rigid deformation.

SOE-Net: A Self-Attention and Orientation Encoding Network for Point Cloud based Place Recognition

1 code implementation CVPR 2021 Yan Xia, Yusheng Xu, Shuang Li, Rui Wang, Juan Du, Daniel Cremers, Uwe Stilla

We tackle the problem of place recognition from point cloud data and introduce a self-attention and orientation encoding network (SOE-Net) that fully explores the relationship between points and incorporates long-range context into point-wise local descriptors.

3D Place Recognition Metric Learning +1

MonoRec: Semi-Supervised Dense Reconstruction in Dynamic Environments from a Single Moving Camera

1 code implementation CVPR 2021 Felix Wimbauer, Nan Yang, Lukas von Stumberg, Niclas Zeller, Daniel Cremers

Unlike other multi-view stereo methods, MonoRec is able to reconstruct both static and moving objects by leveraging the predicted masks.

Deep Shells: Unsupervised Shape Correspondence with Optimal Transport

1 code implementation NeurIPS 2020 Marvin Eisenberger, Aysim Toker, Laura Leal-Taixé, Daniel Cremers

We propose a novel unsupervised learning approach to 3D shape correspondence that builds a multiscale matching pipeline into a deep neural network.

Unsupervised Dense Shape Correspondence using Heat Kernels

no code implementations23 Oct 2020 Mehmet Aygün, Zorah Lähner, Daniel Cremers

In this work, we propose an unsupervised method for learning dense correspondences between shapes using a recent deep functional map framework.

MOTChallenge: A Benchmark for Single-Camera Multiple Target Tracking

no code implementations15 Oct 2020 Patrick Dendorfer, Aljoša Ošep, Anton Milan, Konrad Schindler, Daniel Cremers, Ian Reid, Stefan Roth, Laura Leal-Taixé

We present MOTChallenge, a benchmark for single-camera Multiple Object Tracking (MOT) launched in late 2014, to collect existing and new data, and create a framework for the standardized evaluation of multiple object tracking methods.

Multiple Object Tracking Multiple People Tracking +3

LM-Reloc: Levenberg-Marquardt Based Direct Visual Relocalization

no code implementations13 Oct 2020 Lukas von Stumberg, Patrick Wenzel, Nan Yang, Daniel Cremers

The learned features significantly improve the robustness of direct image alignment, especially for relocalization across different conditions.

Pose Estimation

A Chain Graph Interpretation of Real-World Neural Networks

1 code implementation30 Jun 2020 Yuesong Shen, Daniel Cremers

It is thus a promising framework that deepens our understanding of neural networks and provides a coherent theoretical formulation for future deep learning research.

Effective Version Space Reduction for Convolutional Neural Networks

no code implementations22 Jun 2020 Jiayu Liu, Ioannis Chiotellis, Rudolph Triebel, Daniel Cremers

In active learning, sampling bias could pose a serious inconsistency problem and hinder the algorithm from finding the optimal hypothesis.

Active Learning Image Classification

PrimiTect: Fast Continuous Hough Voting for Primitive Detection

1 code implementation15 May 2020 Christiane Sommer, Yumin Sun, Erik Bylow, Daniel Cremers

This paper tackles the problem of data abstraction in the context of 3D point sets.

Hamiltonian Dynamics for Real-World Shape Interpolation

1 code implementation ECCV 2020 Marvin Eisenberger, Daniel Cremers

While most prior work focuses on synthetic input shapes, our formulation is designed to be applicable to real-world scans with imperfect input correspondences and various types of noise.

Local Distortion

MOT20: A benchmark for multi object tracking in crowded scenes

1 code implementation19 Mar 2020 Patrick Dendorfer, Hamid Rezatofighi, Anton Milan, Javen Shi, Daniel Cremers, Ian Reid, Stefan Roth, Konrad Schindler, Laura Leal-Taixé

The benchmark for Multiple Object Tracking, MOTChallenge, was launched with the goal to establish a standardized evaluation of multiple object tracking methods.

Multi-Object Tracking Multiple Object Tracking with Transformer +2

D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry

no code implementations CVPR 2020 Nan Yang, Lukas von Stumberg, Rui Wang, Daniel Cremers

We propose D3VO as a novel framework for monocular visual odometry that exploits deep networks on three levels -- deep depth, pose and uncertainty estimation.

Monocular Depth Estimation Monocular Visual Odometry

Optimization of Graph Total Variation via Active-Set-based Combinatorial Reconditioning

1 code implementation27 Feb 2020 Zhenzhang Ye, Thomas Möllenhoff, Tao Wu, Daniel Cremers

Structured convex optimization on weighted graphs finds numerous applications in machine learning and computer vision.

Learn to Predict Sets Using Feed-Forward Neural Networks

no code implementations30 Jan 2020 Hamid Rezatofighi, Tianyu Zhu, Roman Kaskman, Farbod T. Motlagh, Qinfeng Shi, Anton Milan, Daniel Cremers, Laura Leal-Taixé, Ian Reid

In our formulation we define a likelihood for a set distribution represented by a) two discrete distributions defining the set cardinally and permutation variables, and b) a joint distribution over set elements with a fixed cardinality.

Multi-Label Image Classification object-detection +1

From Planes to Corners: Multi-Purpose Primitive Detection in Unorganized 3D Point Clouds

2 code implementations21 Jan 2020 Christiane Sommer, Yumin Sun, Leonidas Guibas, Daniel Cremers, Tolga Birdal

We propose a new method for segmentation-free joint estimation of orthogonal planes, their intersection lines, relationship graph and corners lying at the intersection of three orthogonal planes.

Inferring Super-Resolution Depth from a Moving Light-Source Enhanced RGB-D Sensor: A Variational Approach

1 code implementation13 Dec 2019 Lu Sang, Bjoern Haefner, Daniel Cremers

A novel approach towards depth map super-resolution using multi-view uncalibrated photometric stereo is presented.

Depth Map Super-Resolution

Informative GANs via Structured Regularization of Optimal Transport

no code implementations4 Dec 2019 Pierre Bréchet, Tao Wu, Thomas Möllenhoff, Daniel Cremers

We tackle the challenge of disentangled representation learning in generative adversarial networks (GANs) from the perspective of regularized optimal transport (OT).

Representation Learning

Efficient Derivative Computation for Cumulative B-Splines on Lie Groups

6 code implementations CVPR 2020 Christiane Sommer, Vladyslav Usenko, David Schubert, Nikolaus Demmel, Daniel Cremers

Continuous-time trajectory representation has recently gained popularity for tasks where the fusion of high-frame-rate sensors and multiple unsynchronized devices is required.

On the well-posedness of uncalibrated photometric stereo under general lighting

no code implementations17 Nov 2019 Mohammed Brahimi, Yvain Quéau, Bjoern Haefner, Daniel Cremers

While the theoretical foundations of this inverse problem under directional lighting are well-established, there is a lack of mathematical evidence for the uniqueness of a solution under general lighting.

Rolling-Shutter Modelling for Direct Visual-Inertial Odometry

no code implementations4 Nov 2019 David Schubert, Nikolaus Demmel, Lukas von Stumberg, Vladyslav Usenko, Daniel Cremers

The visual part of the system performs a photometric bundle adjustment on a sparse set of points.

Multi-Frame GAN: Image Enhancement for Stereo Visual Odometry in Low Light

no code implementations15 Oct 2019 Eunah Jung, Nan Yang, Daniel Cremers

We propose the concept of a multi-frame GAN (MFGAN) and demonstrate its potential as an image sequence enhancement for stereo visual odometry in low light conditions.

Image Enhancement Optical Flow Estimation +2

Bregman Proximal Framework for Deep Linear Neural Networks

no code implementations8 Oct 2019 Mahesh Chandra Mukkamala, Felix Westerkamp, Emanuel Laude, Daniel Cremers, Peter Ochs

This initiated the development of the Bregman proximal gradient (BPG) algorithm and an inertial variant (momentum based) CoCaIn BPG, which however rely on problem dependent Bregman distances.

Sparse Surface Constraints for Combining Physics-based Elasticity Simulation and Correspondence-Free Object Reconstruction

1 code implementation4 Oct 2019 Sebastian Weiss, Robert Maier, Rüdiger Westermann, Daniel Cremers, Nils Thuerey

In a number of tests using synthetic datasets and real-world measurements, we analyse the robustness of our approach and the convergence behavior of the numerical optimization scheme.

Graphics I.6

Towards Generalizing Sensorimotor Control Across Weather Conditions

no code implementations25 Jul 2019 Qadeer Khan, Patrick Wenzel, Daniel Cremers, Laura Leal-Taixé

The ability of deep learning models to generalize well across different scenarios depends primarily on the quality and quantity of annotated data.

Image-to-Image Translation Translation

Smooth Shells: Multi-Scale Shape Registration with Functional Maps

1 code implementation CVPR 2020 Marvin Eisenberger, Zorah Lähner, Daniel Cremers

Smooth shells define a series of coarse-to-fine shape approximations designed to work well with multiscale algorithms.

Flat Metric Minimization with Applications in Generative Modeling

1 code implementation12 May 2019 Thomas Möllenhoff, Daniel Cremers

We take the novel perspective to view data not as a probability distribution but rather as a current.

Learning to Evolve

1 code implementation8 May 2019 Jan Schuchardt, Vladimir Golkov, Daniel Cremers

Here we show that learning to evolve, i. e. learning to mutate and recombine better than at random, improves the result of evolution in terms of fitness increase per generation and even in terms of attainable fitness.

Evolutionary Algorithms reinforcement-learning +1

GN-Net: The Gauss-Newton Loss for Multi-Weather Relocalization

1 code implementation26 Apr 2019 Lukas von Stumberg, Patrick Wenzel, Qadeer Khan, Daniel Cremers

Direct SLAM methods have shown exceptional performance on odometry tasks.

Visual-Inertial Mapping with Non-Linear Factor Recovery

7 code implementations13 Apr 2019 Vladyslav Usenko, Nikolaus Demmel, David Schubert, Jörg Stückler, Daniel Cremers

We reconstruct a set of non-linear factors that make an optimal approximation of the information on the trajectory accumulated by VIO.

Motion Estimation

Variational Uncalibrated Photometric Stereo under General Lighting

1 code implementation ICCV 2019 Bjoern Haefner, Zhenzhang Ye, Maolin Gao, Tao Wu, Yvain Quéau, Daniel Cremers

Photometric stereo (PS) techniques nowadays remain constrained to an ideal laboratory setup where modeling and calibration of lighting is amenable.

Controlling Neural Networks via Energy Dissipation

no code implementations ICCV 2019 Michael Moeller, Thomas Möllenhoff, Daniel Cremers

The last decade has shown a tremendous success in solving various computer vision problems with the help of deep learning techniques.

Computed Tomography (CT) Deblurring +2

Optimization of Inf-Convolution Regularized Nonconvex Composite Problems

no code implementations27 Mar 2019 Emanuel Laude, Tao Wu, Daniel Cremers

In this work, we consider nonconvex composite problems that involve inf-convolution with a Legendre function, which gives rise to an anisotropic generalization of the proximal mapping and Moreau-envelope.

Homogeneous Linear Inequality Constraints for Neural Network Activations

1 code implementation5 Feb 2019 Thomas Frerix, Matthias Nießner, Daniel Cremers

One way to achieve this task is by means of a projection step at test time after unconstrained training.

Photometric Depth Super-Resolution

1 code implementation26 Sep 2018 Bjoern Haefner, Songyou Peng, Alok Verma, Yvain Quéau, Daniel Cremers

This study explores the use of photometric techniques (shape-from-shading and uncalibrated photometric stereo) for upsampling the low-resolution depth map from an RGB-D sensor to the higher resolution of the companion RGB image.


DeepWrinkles: Accurate and Realistic Clothing Modeling

no code implementations ECCV 2018 Zorah Laehner, Daniel Cremers, Tony Tung

We present a novel method to generate accurate and realistic clothing deformation from real data capture.

Generative Adversarial Network

Detailed Dense Inference with Convolutional Neural Networks via Discrete Wavelet Transform

no code implementations6 Aug 2018 Lingni Ma, Jörg Stückler, Tao Wu, Daniel Cremers

Dense pixelwise prediction such as semantic segmentation is an up-to-date challenge for deep convolutional neural networks (CNNs).

Decoder Semantic Segmentation

LDSO: Direct Sparse Odometry with Loop Closure

no code implementations3 Aug 2018 Xiang Gao, Rui Wang, Nikolaus Demmel, Daniel Cremers

In this paper we present an extension of Direct Sparse Odometry (DSO) to a monocular visual SLAM system with loop closure detection and pose-graph optimization (LDSO).

Loop Closure Detection Translation

Direct Sparse Odometry with Rolling Shutter

no code implementations ECCV 2018 David Schubert, Nikolaus Demmel, Vladyslav Usenko, Jörg Stückler, Daniel Cremers

Neglecting the effects of rolling-shutter cameras for visual odometry (VO) severely degrades accuracy and robustness.

Visual Odometry

The Double Sphere Camera Model

9 code implementations24 Jul 2018 Vladyslav Usenko, Nikolaus Demmel, Daniel Cremers

We evaluate the model using a calibration dataset with several different lenses and compare the models using the metrics that are relevant for Visual Odometry, i. e., reprojection error, as well as computation time for projection and unprojection functions and their Jacobians.

3D Reconstruction Autonomous Driving +3

Modular Vehicle Control for Transferring Semantic Information Between Weather Conditions Using GANs

1 code implementation3 Jul 2018 Patrick Wenzel, Qadeer Khan, Daniel Cremers, Laura Leal-Taixé

To this end, we propose to divide the task of vehicle control into two independent modules: a control module which is only trained on one weather condition for which labeled steering data is available, and a perception module which is used as an interface between new weather conditions and the fixed control module.

Generative Adversarial Network Self-Driving Cars

Divergence-Free Shape Interpolation and Correspondence

1 code implementation27 Jun 2018 Marvin Eisenberger, Zorah Lähner, Daniel Cremers

We present a novel method to model and calculate deformation fields between shapes embedded in $\mathbb{R}^D$.

q-Space Novelty Detection with Variational Autoencoders

1 code implementation8 Jun 2018 Aleksei Vasilev, Vladimir Golkov, Marc Meissner, Ilona Lipp, Eleonora Sgarlata, Valentina Tomassini, Derek K. Jones, Daniel Cremers

Since abnormal samples are not used during training, we define novelty metrics based on the (partially complementary) assumptions that the VAE is less capable of reconstructing abnormal samples well; that abnormal samples more strongly violate the VAE regularizer; and that abnormal samples differ from normal samples not only in input-feature space, but also in the VAE latent space and VAE output.

Novelty Detection

Fight Ill-Posedness With Ill-Posedness: Single-Shot Variational Depth Super-Resolution From Shading

1 code implementation CVPR 2018 Bjoern Haefner, Yvain Quéau, Thomas Möllenhoff, Daniel Cremers

We put forward a principled variational approach for up-sampling a single depth map to the resolution of the companion color image provided by an RGB-D sensor.


Deep Perm-Set Net: Learn to predict sets with unknown permutation and cardinality using deep neural networks

no code implementations ICLR 2019 S. Hamid Rezatofighi, Roman Kaskman, Farbod T. Motlagh, Qinfeng Shi, Daniel Cremers, Laura Leal-Taixé, Ian Reid

We demonstrate the validity of this new formulation on two relevant vision problems: object detection, for which our formulation outperforms state-of-the-art detectors such as Faster R-CNN and YOLO, and a complex CAPTCHA test, where we observe that, surprisingly, our set based network acquired the ability of mimicking arithmetics without any rules being coded.

object-detection Object Detection

The TUM VI Benchmark for Evaluating Visual-Inertial Odometry

4 code implementations17 Apr 2018 David Schubert, Thore Goll, Nikolaus Demmel, Vladyslav Usenko, Jörg Stückler, Daniel Cremers

For trajectory evaluation, we also provide accurate pose ground truth from a motion capture system at high frequency (120 Hz) at the start and end of the sequences which we accurately aligned with the camera and IMU measurements.

Visual Odometry

Direct Sparse Visual-Inertial Odometry using Dynamic Marginalization

1 code implementation16 Apr 2018 Lukas von Stumberg, Vladyslav Usenko, Daniel Cremers

We present VI-DSO, a novel approach for visual-inertial odometry, which jointly estimates camera poses and sparse scene geometry by minimizing photometric and IMU measurement errors in a combined energy functional.

Clustering with Deep Learning: Taxonomy and New Methods

2 code implementations23 Jan 2018 Elie Aljalbout, Vladimir Golkov, Yawar Siddiqui, Maximilian Strobel, Daniel Cremers

In this paper, we propose a systematic taxonomy of clustering methods that utilize deep neural networks.


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

Combinatorial Preconditioners for Proximal Algorithms on Graphs

no code implementations16 Jan 2018 Thomas Möllenhoff, Zhenzhang Ye, Tao Wu, Daniel Cremers

We present a novel preconditioning technique for proximal optimization methods that relies on graph algorithms to construct effective preconditioners.

BIG-bench Machine Learning

Regularization for Deep Learning: A Taxonomy

no code implementations ICLR 2018 Jan Kukačka, Vladimir Golkov, Daniel Cremers

Regularization is one of the crucial ingredients of deep learning, yet the term regularization has various definitions, and regularization methods are often studied separately from each other.

Online Photometric Calibration for Auto Exposure Video for Realtime Visual Odometry and SLAM

no code implementations5 Oct 2017 Paul Bergmann, Rui Wang, Daniel Cremers

We further show that our calibration can improve the performance of a state-of-the-art direct visual odometry method that works solely on pixel intensities, calibrating for photometric parameters in an online fashion in realtime.

Camera Calibration Visual Odometry

Variational Reflectance Estimation from Multi-view Images

no code implementations25 Sep 2017 Jean Mélou, Yvain Quéau, Jean-Denis Durou, Fabien Castan, Daniel Cremers

We tackle the problem of reflectance estimation from a set of multi-view images, assuming known geometry.

Video Object Segmentation Without Temporal Information

no code implementations18 Sep 2017 Kevis-Kokitsi Maninis, Sergi Caelles, Yu-Hua Chen, Jordi Pont-Tuset, Laura Leal-Taixé, Daniel Cremers, Luc van Gool

Video Object Segmentation, and video processing in general, has been historically dominated by methods that rely on the temporal consistency and redundancy in consecutive video frames.

Foreground Segmentation Object +5

Efficient Online Surface Correction for Real-time Large-Scale 3D Reconstruction

no code implementations12 Sep 2017 Robert Maier, Raphael Schaller, Daniel Cremers

State-of-the-art methods for large-scale 3D reconstruction from RGB-D sensors usually reduce drift in camera tracking by globally optimizing the estimated camera poses in real-time without simultaneously updating the reconstructed surface on pose changes.

3D Reconstruction

Stereo DSO: Large-Scale Direct Sparse Visual Odometry with Stereo Cameras

no code implementations ICCV 2017 Rui Wang, Martin Schwörer, Daniel Cremers

We propose Stereo Direct Sparse Odometry (Stereo DSO) as a novel method for highly accurate real-time visual odometry estimation of large-scale environments from stereo cameras.

3D Reconstruction Optical Flow Estimation +1