DAGM GCPR 2021
DAGM German Conference on Pattern Recognition, September 28 - October 1, 2021
DAGM German Conference on Pattern Recognition, September 28 - October 1, 2021
Welcome to the 43rd DAGM German Conference on Pattern Recognition, the annual symposium of the German Association for Pattern Recognition (DAGM). The conference is an international premier venue for recent advances in pattern recognition including image processing, machine learning, and computer vision and welcomes submissions from all areas of pattern recognition. It features special tracks, nectar tracks, a special session on Unsolved Problems in Pattern Recognition, the German Pattern Recognition Award for outstanding research in the fields of pattern recognition, computer vision, and machine learning, the DAGM MVTec Dissertation Award, and the Young Researchers' Forum with an award for the best Master thesis. The best papers will be invited to contribute to a special issue of the International Journal of Computer Vision (IJCV).
Thanks to all presenters, reviewers, session chairs, supporting staff, and over 720 registered participants. As registered participant, you can still access the PDFs until end of October via virtual21.dagm-gcpr.de. The recorded talks are available at the DAGM GCPR 2021 YouTube Channel.
The preliminary program is available and includes four keynotes by David Forsyth (University of Illinois at Urbana-Champaign), Kristen Grauman (University of Texas at Austin, Facebook AI Research), Thorsten Joachims (Cornell University), and Jiri Matas (Czech Technical University, Prague).
A tutorial on Geometric Deep Learning will be given by Emanuele Rodolà (Sapienza University of Rome) and a workshop on Scene Understanding in Unstructured Environments will be organized on September 28. The day will be complemented by a Machine Learning Nectar Track and a Pattern Recognition and Computer Vision Nectar Track.
DAGM GCPR 2021 Best Paper Award
InfoSeg: Unsupervised Semantic Image Segmentation with Mutual Information Maximization
Robert Harb (TU Graz) and Patrick Knöbelreiter (TU Graz)
DAGM GCPR 2021 Honorable Mentions
TxT: Crossmodal End-to-End Learning with Transformers
Jan-Martin O. Steitz (TU Darmstadt), Jonas Pfeiffer (TU Darmstadt), Iryna Gurevych (TU Darmstadt), and Stefan Roth (TU Darmstadt)
Video Instance Segmentation with Recurrent Graph Neural Networks
Joakim Johnander (Linköping University), Emil Brissman (Linköping University), Martin Danelljan (ETH Zurich), and Michael Felsberg (Linköping University)
German Pattern Recognition Award
Zeynep Akata (University of Tübingen)
DAGM MVTec Dissertation Award
Stability and Expressiveness of Deep Generative Models
Lars Morten Mescheder (University of Tübingen)
DAGM Best Master's Thesis Award
(SP)²Net for Generalized Zero-Label Semantic Segmentation
Anurag Das (Saarland University)
If you have questions, please contact us per email to dagm-gcpr@googlegroups.com.
The DAGM German Conference on Pattern Recognition (DAGM GCPR) 2021 is the 43rd annual symposium of the German Association for Pattern Recognition (DAGM). It is an international premier venue for recent advances in pattern recognition including image processing, machine learning, and computer vision and welcomes submissions from all areas of pattern recognition. Authors are invited to submit high-quality papers presenting original research. Submitted papers will be reviewed based on the criteria of originality, soundness, empirical evaluation, and presentation. Accepted papers will be published by Springer as a proceeding of the Lecture Notes in Computer Science (LNCS). The best papers will be invited to contribute to a special issue of the International Journal of Computer Vision (IJCV). Revised ICCV 2021 submissions can be submitted to the Fast Review Track.
Topics of interest include, but are not limited to, the following:
We especially invite submissions for these Special Tracks, which are chaired and reviewed explicitly by experts from the respective fields:
Computer vision systems and applications
Chairs: Bodo Rosenhahn (Leibniz University Hannover),
Carsten Steger (MVTec Software GmbH, Technical University of Munich)
The computer vision systems and applications track invites papers on systems and applications with significant, interesting vision and machine learning components. The track provides a forum for researchers working on industrial applications to share their latest developments. The reviewing criteria will be slightly different for this track: The focus is not on state-of-the-art research novelties, but the system and applied papers have to stand out in the successful transfer and application of research results to industry with measurable success indicators, such as performance, robustness, memory or energy consumption, big data, the systems-level innovation or the adaptation of existing methods to a complete novel domain while satisfying industrial requirements. These review criteria will be explicitly communicated to the reviewers to ensure clear quality expectations and interesting contributions.
Pattern recognition in the life and natural sciences
Chairs: Joachim Denzler (University of Jena), Xiaoyi Jiang (University of Münster)
Pattern recognition and machine learning already became a major driver in the sciences, for example, for data driven analysis or understanding of processes. This special track asks for original work that demonstrates successful development and application of pattern recognition methods tailored to the specific domain from the life- and natural sciences.
Photogrammetry and remote sensing
Chairs: Helmut Mayer (Bundeswehr University Munich), Uwe Sörgel (University of Stuttgart)
The photogrammetry and remote sensing track invites papers on theory and applications in photogrammetry and remote sensing with significant computer vision or machine learning components. The track provides a forum for researchers developing approaches from image classification and segmentation to high-precision photogrammetry to share their latest developments. The reviewing criteria will be slightly different for this track: Besides based on state-of-the-art research novelties papers will also be considered if they present interesting, complex applications possibly in unexpected domains or with novel extensive data sets. These review criteria will be explicitly communicated to the reviewers to ensure clear quality expectations and interesting contributions.
Robot vision
Chairs: Friedrich Fraundorfer (Graz University of Technology), Jörg Stückler (Max Planck Institute for Intelligent Systems)
The robot vision track invites papers on state-of-the-art research in computer vision approaches for robotics. The papers in the track will be reviewed by experts in the field and judged by criteria of technical merit, quality, originality, and scientific novelty. The track provides a forum for researchers on robotics-related methods for computer vision and machine learning at the conference.
Papers are submitted through Microsoft CMT (https://cmt3.research.microsoft.com/DAGMGCPR2021) using the author kit.The dates for all submissions except of the Fast Review Track (deadlines will not be extended):
The dates for the Fast Review Track (deadlines will not be extended):
Please read carefully the Camera Ready Instructions. In order to have enough space to add authors, acknowledgments, and to address the comments of the reviewers, you have 13 pages excluding references (1 page more than for the submitted version). The paper will only be accepted if all steps will be completed until 9.9. anywhere on earth:
All files are submitted through Microsoft CMT (https://cmt3.research.microsoft.com/DAGMGCPR2021) using the author kit. See also the sample PDF from the author kit.
The DAGM GCPR 2021 proceedings to be published in the Springer LNCS series will include all accepted papers under the condition that a paper processing fee is paid for each accepted paper. Authors should take into account the following rules:
Papers that do not conform with the above guidelines will be rejected without reviewing.
If you are a Master student you might be eligible for the Best Master's Thesis Award of the Young Researchers Forum. Students who received a Master's degree from a university in Austria, Germany, or Switzerland after July 10, 2020 can apply for the best Master's Thesis Award. Please visit the DAGM Young Researcher Forum website for more information. The submission instructions are the same as for the Fast Review Track, but the following steps need to be done in addition:
Send an email with subject YRF Fast Review Track Submission: #PaperID to dagm-gcpr@googlegroups.com before the submission deadline. The email should state that #PaperID is a YRF submission and contain a link for downloading the additional documents:
Please do not include these documents in the supplementary material submitted via CMT since it reveals the identity of the author.
In the LaTeX template, please change "\def\GCPRTrack{Fast Review Track}" to "\def\GCPRTrack{YRF Fast Review Track}".
If your ICCV 2021 submission has not been accepted but you can address the concerns of the ICCV reviewers by a minor revision of the paper, you can submit a revised version of the paper to the Fast Review Track. The review process takes only 3 weeks and is similar to the reviewing process of a journal for a minor revision. For preparing a submission, please use the author kit which already defines "\def\GCPRTrack{Fast Review Track}". As for regular papers, the paper length is limited to 12 pages excluding references. It is not allowed to modify the margins, font size, or page layout of the template. Submissions that use the ICCV template will be rejected. Besides any optional supplementary material for the revised paper, the following documents need to be provided as part of the supplementary material:
In order to get the ICCV reviews and meta-review, print them to PDF using CMT. For the reviews, click on "View Reviews". You will see the title of the ICCV submission, the paper ID of the ICCV submission, and all reviews. At the top right, you have to click on "Print", select Print to PDF as printer (name depends on the installed PDF printer), and print the reviews to the file reviews.pdf. For the meta-review, click on "View Meta-Reviews" and print it in the same way to the file meta-review.pdf. The summary of changes needs to describe how and where in the revised paper or supplementary material the points of the reviewers have been addressed. Please structure the response by reviewer and go through each point of the reviews. Please ensure that the submission including any supplementary material does not contain any information that may identify the authors. See policies. There are no specific formatting requirements for the summary of changes, but you can use the Latex template.
By submitting a paper to the Fast Review Track, the authors agree that the submission can be shared with the organizers of ICCV 2021 in order to verify the original ICCV submission and reviews.
Dates for Fast Review Track (deadlines will not be extended):
The goals of DAGM GCPR are to publish exciting new work for the first time and to avoid duplicating the effort of reviewers.
By submitting a manuscript to DAGM GCPR, authors acknowledge that it has not been previously published or accepted for publication in substantially similar form in any peer-reviewed venue including journal, conference or workshop, or archival forum. Furthermore, no publication substantially similar in content has been or will be submitted to this or another conference, workshop, or journal during the review period. Violation of any of these conditions will lead to rejection, and will be reported to the other venue to which the submission was sent.
A publication, for the purposes of this policy, does not consider an arXiv.org paper as a publication because it cannot be rejected. It also excludes technical reports which are not peer reviewed.
By submitting a paper, the authors agree that the paper is processed by the Toronto Paper Matching System to match each manuscript to the best possible reviewers.
If you have questions, please contact us per email to dagm-gcpr@googlegroups.com.
For preparing a submission, please use the author kit. If you use overleaf, please use dagmgcpr2021.tex from the author kit as well and add the submission id, which you will receive from Microsoft CMT after creating a new submission, and the submission track. See also the sample PDF from the author kit.
Papers are submitted through Microsoft CMT (https://cmt3.research.microsoft.com/DAGMGCPR2021).
We invite to a Special Session on "Unsolved Problems in Pattern Recognition" as part of the DAGM German Conference on Pattern Recognition (DAGM GCPR) 2021. The session provides a platform for discussing the major challenges of pattern recognition, computer vision, and machine learning in the next years. The event is an opportunity to take a step back from the daily business and debate about the currently most relevant problems in the field and emphasize the most promising future research directions. Participants are kindly requested to submit an extended abstract of not more than two pages, which sketches an open problem in the respective field or in related application areas. The proposal should contain:
A committee will select a small number of proposals to be presented in talks. The goal of the talks is to spark a (possibly controversial) discussion rather than presenting paper-ready solutions to the depicted problem. Please submit your proposal as single PDF per email to dagm-gcpr@googlegroups.com. If you have questions, please contact us per email as well.
Keynote speakers of this Special Session are David Forsyth and Jiri Matas. The list of speakers is available in the program.We invite to present high-quality published papers on machine learning and AI that have been presented at other conferences or journals in the last two years. Examples for conferences of interest include AAAI, ACL, ICLR, ICML, IJCAI, NeurIPS, UAI.
We invite to present high-quality published papers on pattern recognition and computer vision that have been presented at other conferences or journals in the last two years. Examples for journals or conferences of interest include CVPR, ECCV, ICASSP, ICCV, IJCV, MICCAI, TIP, TPAMI.
Participants are kindly requested to send an email with subject "Machine Learning Nectar Track" or "Pattern Recognition and Computer Vision Nectar Track" to dagm-gcpr@googlegroups.com. The email should contain:
A committee will select the presentations based on scientific quality, relevance, and available time slots. If you have questions, please contact us per email as well.
The conference will be held from September 28 to October 1, 2021. The conference is organized as a virtual conference. The participation is free of charge, but a registration is required: virtual21.dagm-gcpr.de. The website will give you access to all sessions, PDFs of accepted papers, and talks.
On September 28, there will be a tutorial on Geometric Deep Learning, a workshop on Scene Understanding in Unstructured Environments, and two Nectar Tracks for Machine Learning and Pattern Recognition. The Nectar Tracks offer the opportunity to present and discuss the latest works that have been published at top-tier machine learning or pattern recognition conferences and journals. The special session on Unsolved Problems in Pattern Recognition on September 29 provides a unique venue for discussing the major challenges of pattern recognition in the next years. The event is an opportunity to take a step back from the daily business and debate about the currently most relevant problems in the field and emphasize the most promising future research directions.
The workshop is organized by Abhinav Valada (University of Freiburg), Peter Mortimer (Bundeswehr University Munich), Nina Heide (Fraunhofer IOSB), and Jens Behley (University of Bonn). The workshop includes a challenge on Outdoor Semantic Segmentation. For more details including program and submission deadlines for the challenge as well as extended abstracts please visit Workshop on Scene Understanding in Unstructured Environments. The workshop also includes invited talks by:
The past decade in computer vision research has witnessed the re-emergence of "deep learning", and in particular convolutional neural network (CNN) techniques, allowing to learn powerful image feature representations from large collections of examples. CNNs achieve a breakthrough in performance in a wide range of applications such as image classification, segmentation, detection and annotation. Nevertheless, when attempting to apply the CNN paradigm to 3D shapes (feature-based description, similarity, correspondence, retrieval, etc.) one has to face fundamental differences between images and geometric objects. Shape analysis and geometry processing pose new challenges that are non-existent in image analysis, and deep learning methods have only recently started penetrating into our community. The purpose of this tutorial is to overview the foundations and the current state of the art on learning techniques for 3D shape analysis. Special focus will be put on deep learning techniques (CNN) applied to Euclidean and non-Euclidean manifolds for tasks of shape classification, retrieval and correspondence. The tutorial will present in a new light the problems of shape analysis and geometry processing, emphasizing the analogies and differences with the classical 2D setting, and showing how to adapt popular learning schemes in order to deal with 3D shapes. The tutorial will assume no particular background, beyond some basic working knowledge that is a common denominator for students and practitioners in vision and graphics. The recording of the second part is available at: video.
The conference will be held as a virtual conference, but you are welcome to visit Bonn at another time.
Bonn is a green city in the heart of Europe, beautifully situated on the banks of the river Rhine. With its idyllic green surroundings, it waits to be explored by nature-lovers, hikers and cyclists. Here, at the gateway to the Upper Middle Rhine Valley - a UNESCO World Heritage site - visitors encounter a hospitable place with rich history, many cultural offerings, a cosmopolitan flair and open-minded citizens. Bonn is a multicultural city where people from around 180 nationalities live peacefully together. The city features a rich history and a great number of cultural events throughout the year. The Museum Mile and several other museums, including the birthplace of Ludwig van Beethoven, are located in and around the city center. Visitors come from all around the world to enjoy various events and festivals and to go to the theatre and the opera. An overview of places of interest can be found at bonn-region.de. The following videos give some authentic impressions of Bonn:
ZEISS is optics and innovation. Our over 32,000 colleagues develop, manufacture and sell highly innovative products and solutions for our customers in a variety of business fields generating revenue of over €6 billion.
Vzense founded in 2016, a professional TOF 3D sensor and system provider, focus on AMR, logistics, people counting and many customized TOF applications for our customers in the United States, EU, Japan and China. Contact: info@vzense.com.
Prof. Dr. Juergen Gall
gall@iai.uni-bonn.de, Phone: +49 228 73 69600
Prof. Dr. Juergen Gall
Department of Information Systems and Artificial Intelligence
Friedrich-Hirzebruch-Allee 8, 53115 Bonn, Germany
gall@iai.uni-bonn.de, Phone: +49 228 73 69600
Die auf den Web-Seiten abrufbaren Beiträge dienen nur der allgemeinen Information und nicht der Beratung in konkreten Fällen. Wir sind bemüht, für die Richtigkeit und Aktualität aller auf der Website enthaltenen Informationen und Daten gemäß § 7 Abs.1 TMG zu sorgen. Für die Korrektheit, die Vollständigkeit, die Aktualität oder Qualität der bereitgestellten Informationen und Daten wird jedoch keine Gewähr nach § 8 bis 10 TMG übernommen. Die Haftung für den Inhalt der abrufbaren Informationen wird ausgeschlossen, soweit es sich nicht um vorsätzliche oder grob fahrlässige Falschinformation handelt. Verpflichtungen zur Entfernung oder Sperrung der Nutzung von Informationen nach den allgemeinen Gesetzen bleiben hiervon unberührt. Eine diesbezügliche Haftung ist jedoch erst ab dem Zeitpunkt der Kenntnis einer konkreten Rechtsverletzung möglich. Bei Bekanntwerden von entsprechenden Rechtsverletzungen werden wir diese Inhalte umgehend entfernen.
Wir sind für den Inhalt von Webseiten, die über einen Hyperlink erreicht werden, nicht verantwortlich. Für den Inhalt der verlinkten Seiten sind ausschließlich deren Betreiber verantwortlich. Wir machen uns die Inhalte dieser Internetseiten ausdrücklich nicht zu eigen und können deshalb für die inhaltliche Korrektheit, Vollständigkeit und Verfügbarkeit keine Gewähr leisten. Wir haben bei der erstmaligen Verknüpfung zwar den fremden Inhalt daraufhin überprüft, ob durch ihn eine mögliche zivilrechtliche oder strafrechtliche Verantwortlichkeit ausgelöst wird. Wir sind aber nicht dazu verpflichtet, die Inhalte, auf die wir unserem Angebot verweisen, ständig auf Veränderungen zu überprüfen, die eine Verantwortlichkeit neu begründen könnten. Erst wenn wir feststellen oder von anderen darauf hingewiesen werden, dass ein konkretes Angebot, zu dem wir einen Link bereitgestellt haben, eine zivil- oder strafrechtliche Verantwortlichkeit auslöst, werden wir den Verweis auf dieses Angebot aufheben, soweit uns dies technisch möglich und zumutbar ist.
Die durch den Betreiber dieser Seite erstellten Inhalte und Werke auf diesen Webseiten unterliegen dem deutschen Urheberrecht. Sämtliche Beiträge Dritter sind als solche gekennzeichnet. Die Vervielfältigung, Bearbeitung, Verbreitung und jede Art der Verwertung außerhalb der Grenzen des Urheberrechts bedürfen der schriftlichen Zustimmung des jeweiligen Autors bzw. Erstellers. Kopien von diesen Seiten sind nur für den privaten Bereich gestattet, nicht jedoch für kommerzielle Zwecke.
Die Nutzung unserer Webseite ist in der Regel ohne Angabe personenbezogener Daten möglich. Soweit auf unseren Seiten personenbezogene Daten (beispielsweise Name, Anschrift oder E-Mail-Adressen) erhoben werden, erfolgt dies, soweit möglich, stets auf freiwilliger Basis. Diese Daten werden ohne Ihre ausdrückliche Zustimmung nicht an Dritte weitergegeben.
In Verbindung mit Ihrem Zugriff werden in unserem Server für Zwecke der Datensicherheit vorübergehend Daten gespeichert, die möglicherweise eine Identifizierung zulassen (zum Beispiel IP-Adresse, Datum, Uhrzeit und Name der angeforderten Datei). Eine Auswertung, mit Ausnahme für statistische Zwecke in anonymisierter Form, erfolgt nicht.
Quelle: Impressumspflicht, Rechtsfragen
Content Registration
Content Jobs
