On May 15^th Dr. Oliver Wasenmüller was giving a Keynote Lecture on visual perception in autonomous driving on the Mannheim Computer Science Colloquium. In his presentation, he introduced the different stages of autonomous driving and explained the central role of visual perception for them. Decisive for automated driving functions is, for example, the determination of distances to other road users (such as vehicles, cyclists, pedestrians, etc.), the detection of lanes and traffic signs or the precise localization of one’s own vehicle. Furthermore, the movements in the environment have to be captured accurately in order to make predictions about the near future. The research of Dr. Wasenmüller’s team ‘Machine Vision and Autonomous Vehicles’ deals exactly with these challenges and is looking for more innovative methods to better perceive the vehicle environment.

The Mannheim Computer Science Colloquium serves to present and discuss new research results as well as new methods, techniques and solutions of computer science. Every month another speaker from science or industry is invited to present the latest trends in a given topic.

Prof. Stricker presented at the INNOSPACE event of Airbus, ESA and HBO the current state of the art of Augmented Reality and newest results about how to automate content creation (see start-up www.ioxp.de) and initialize 3D tracking without any preparation or set-up work.

It is a great honour to announce that our Horizon2020 project LARA, won the Madrid Challenge of the European Satellite Navigation Competition, ESNC 2017. GeoImaging from Nicosia, coordinator and technology partner of LARA project together with SignalGeneriX from Limassol, a technology partner of the project, received the prestigious award on behalf of the LARA consortium.

The ESNC Awards Ceremony took place November 7, 2017 in front of an international audience during the European Space Week 2017 in Tallinn, Estonia.

Dr Stylianidis, coordinator of LARA project and Dr Tasos Kounoudes of SignalGeneriX receive the award from the Director of the GSA, Mr Carlo des Dorides

The ESNC annually awards the best services, products, and business ideas using satellite navigation in everyday life, spurring the development of respective market-driven applications.

This award is a confirmation of success for all the hard work invested by all project partners especially now that the project enters the commercialization phase.

A complete list of this year’s ESNC winners is available at the ESNC 2017 Hal of Fame.

The group Augmented Vision implemented an embedded and ultra-low power machine learning recognition algorithm, which has been tested in the context of trials at the Albertina Museum in Vienna, Austria. The Eyes of Thing is a visual IoT System with AI capability. It has been jointly developed with 8 partners across Europe over the last three years and anticipated the current and very fast development of Embedded Machine Vision using Learning Techniques. High-level computer vision and machine learning are being embedded in more and more devices, paving the way to many new applications.

See the video of the trials, by our partner Fluxguide:

Paper #1 : Fusion of Keypoint Tracking and Facial Landmark Detection for Real-Time Head Pose Estimation by Jilliam Maria Diaz Barros, Bruno Mirbach, Frederic Garcia, Kiran Varanasi, Didier Stricker

Paper #2: SceneFlowFields: Dense Interpolation of Sparse Scene Flow Correspondences by René Schuster, Oliver Wasenmüller, Georg Kuschk, Christian Bailer, Didier Stricker

Paper #3: 3D Shape Processing by Convolutional Denoising Autoencoders on Local Patches by Kripasindhu Sarkar, Kiran Varanasi, Didier Stricker

Please follow this link https://www.dfki.de/web/presse/pressemitteilung/2017/alterego for the German press release.

Check out this and other job opportunities here.

Congratulations to our colleagues for this success!

The department Augmented Vision demonstrates its object tracking technology “Augmented Things” at ISMAR 2017 in Nantes! This technology allows to augment any object for everyday use – the tracking is especially robust in difficult light conditions and under partial occlusion. Besides the demonstrator “Augmented Things”, the department Augmented Vision presents two publications:

Augmented Things: Enhancing AR Applications leveraging the Internet of Things and Universal 3D Object Tracking
Jason Raphael Rambach, Alain Pagani, Didier Stricker

Fusion of unsynchronized optical tracker and inertial sensor in EKF framework for in-car Augmented Reality delay reduction
Jason Raphael Rambach, Alain Pagani, Sebastian Lampe, Ruben Reiser, Manthan Pancholi, Didier Stricker

A Probablistic Combination of CNN and RNN Estimates for Hand Gesture Based Interaction in Car
Aditya Tewari, Bertram Taetz, Frederic Grandidier, Didier Stricker

3DV is a rapidly growing conference focusing on all aspects of 3D vision and 3D processing.

3DV provides “a premier platform for disseminating research results covering a broad variety of topics in the area of 3D research in computer vision and graphics, from novel optical sensors, signal processing, geometric modeling, representation and transmission, to visualization and interaction, and a variety of applications”.

Our following papers has been accepted and will be presented at the conference in Qingdao from October 10th-12th 2017:

Paper  # 1: Simultaneous Hand Pose and Skeleton Bone-Lengths Estimation from a Single Depth Image by Jameel Malik, Ahmed Elhayek, and Didier Stricker.

Paper  # 2: Fast and Efficient Depth Map Estimation from Light Fields by Yuriy Anisimov and Didier Stricker.

Paper  # 3: Learning Quadrangulated Patches for 3D Shape Parameterization and Completion by Kripasindhu Sarkar, Kiran Varanasi, and Didier Stricker.

Paper  # 4: Scalable Dense Monocular Surface Reconstruction by Mohammad Dawud Ansari, Vladislav Golyanik, and Didier Stricker.

Paper  # 5: High Dimensional Space Model for Dense Monocular Surface Recovery by Vladislav Golyanik and Didier Stricker.

Paper  # 6: Multiframe Scene Flow with Piecewise Rigid Motion by Vladislav Golyanik, Kihwan Kim, Robert Maier, Matthias Nießner, Didier Stricker, and Jan Kautz.

The papers will be shortly available on our webpage.

Check our job offers here.

Read the interesting article on Augmented Reality from normal videos published in the online issue on September 11th, 2017:

https://www.newscientist.com/article/2146850-app-creates-augmented-reality-tutorials-from-normal-videos/

On 1^st September 2017, Oliver Wasenmüller successfully defended his PhD thesis about “Towards an Accurate RGB-D Benchmark, Mapping and Odometry as well as their Applications” in front of the examination commission consisting of Prof. Dr. Didier Stricker (TU Kaiserslautern and DFKI), Prof. Dr. Dieter Fellner (TU Darmstadt and FhG IGD) and Prof. Dr. Karsten Berns (TU Kaiserslautern).

In his thesis, Oliver Wasenmüller addressed and enhanced several algorithms for more accurate RGB-D mapping and odometry. For that purpose, he researched methods for the precise mapping of depth images as well as the determination of their positions in odometry algorithms. He provided a public benchmark, which, for the first time, allows the objective evaluation of such algorithms. Furthermore, he transferred the algorithms into an industrial application within the scope of a 3D discrepancy check. He conducted his research in the context of the projects ARVIDA and BodyAnalyzer.

Oliver Wasenmüller studied Computer Science at the University of Kaiserslautern and finished his master in cooperation with the Sony European Technology Center. Since then, he has been working as a project manager and researcher at the department “Augmented Vision” (AV) of the DFKI Kaiserslautern. He is currently managing several projects dealing with image processing for automotive and industrial applications.

Prof. Dr. Stricker announces the successfully passed PhD of Dr. Wasenmüller along with Prof. Dr. Fellner and Prof. Dr. Berns

We are proud to announce that our research department got one of the highly coveted NVIDIA Tesla V100s. NVIDIA CEO, Jensen Huang, not only unveiled NVIDIA’s novelty in the context of the CVPR 2017 but he also presented it to 15 participants of the NVIDIA AI Labs programes including Augmentd Vision department. In the front row, first from right, you can see our colleague, Christian Bailer, who attended the conferece and received the gift on behalf of the group.

Check the details on NVIDIA’s blog.

Our research group got two papers accepted at British Machine Vision Conference (BMVC) 2017:

Fast dense feature extraction with convolutional neural networks that have pooling or striding layers by Christian Bailer, Tewodros Habtegebrial, Kiran Varanasi and Didier Stricker.

Introduction to Coherent Depth Fields for Dense Monocular Surface Recovery by Vladislav Golyanik, Torben Fetzer and Didier Stricker.

Both papers will be soon available on our webpage.

The BMVC is one of the major international conferences on computer vision and related areas. It is organised by the British Machine Vision Association (BMVA) and will take place from 4th to 7th September 2017 in the Imperial College London.

The paper analyses applicability of current GPUs in real-time applications with hard real-time
constraints and provides a solution for a single dedicated GPU. This work was done in collaboration with
Max Planck Institute for Software Systems (MPI-SWS).

Towards Scheduling Hard Real-Time Image Processing Tasks on a Single GPU
Vladislav Golyanik, Mitra Nasri, Didier Stricker

The IEEE International Conference on Image Processing (ICIP) 2017 will feature world-class speakers, tutorials, and industry sessions, and creates an excellent forum to foster innovation and entrepreneurship, and network with the brightest minds in academia and industry that are working in this field. The conference  will be held at the China National Convention Center in Beijing, China from 17-20 September 2017.

Full description see Thesis-Monocular 3DRec

Erweiterte Realität im Mannheimer Morgen

Our paper on deep-learning and optical flow will be presented @ CVPR 2017.

Please find the the details here:

CNN-based Patch Matching for Optical Flow with Thresholded Hinge Embedding Loss

CVPR is the premier annual computer vision event comprising the main conference and several co-located workshops and short courses. With its high quality and low cost, it provides an exceptional value for students, academics and industry researchers.