The Third Workshop and Challenge on Deep Learning for Geometric Computing in conjunction with ICCV 2021

Computer vision approaches have made tremendous efforts toward understanding shape from various data formats, especially since entering the deep learning era. Although accurate results have been obtained in detection, recognition, and segmentation, there is less attention and research on extracting topological and geometric information from shapes. These geometric representations provide compact and intuitive abstractions for modeling, synthesis, compression, matching, and analysis. Extracting such representations is significantly different from segmentation and recognition tasks, as they contain both local and global information about the shape. To attract attention of researchers from computer vision, computational geometry, computer graphics, and machine learning to this branch of problems, we organize the third edition of “Deep Learning for Geometric Computing” workshop at ICCV 2021. The workshop encapsulates competitions with prizes, proceedings, keynotes, paper presentations, and a fair and diverse environment for brainstorming about future research collaborations.

*************** Call for competition participation ***************

We are hosting seven competition tracks in two main domains: The SkelNetOn Challenge (2D) and The ABC Challenge (3D), implemented as independent contests available at Codalab.

*** The SkelNetOn Challenge ***

The SkelNetOn Challenge is structured around shape understanding in four domains. We provide shape datasets and some complementary resources (e.g, pre/post-processing, sampling, and data augmentation scripts) and the testing platform.

Submissions to the challenge will perform one of the following tasks:

– Shape pixels to skeleton pixels https://competitions.codalab.org/competitions/21169

– Shape points to skeleton points https://competitions.codalab.org/competitions/21172

– Shape pixels to parametric curves https://competitions.codalab.org/competitions/21175

– Natural image pixels to skeleton pixels https://competitions.codalab.org/competitions/24536 

*** The ABC Challenge ***

The ABC Challenge serves as a testbed for common shape analysis and geometry processing tasks. We supplement the challenge with additional software libraries, sets of large-scale standardized benchmarks (data splits, resolutions, and targets), and implementations of evaluation metrics. The first ABC challenge will be hosting a three-track contest on geometry processing, including:

– Estimation of non-oriented normals https://competitions.codalab.org/competitions/24253

– Geometric shape segmentation https://competitions.codalab.org/competitions/25087

– Sharpness fields extraction https://competitions.codalab.org/competitions/25079

*************** Call for paper submissions ***************

We will have an open submission format where i) participants in the competition will be required to submit a paper, or ii) researchers can share their novel unpublished research in deep learning for geometric computing. The top submissions in each category will be invited to present their work during the workshop and will be published in the workshop proceedings. The workshop will also honor the best paper and the best student paper.

Although we encourage all submissions to benchmark their results on the evaluation platform, there are other relevant research areas that our datasets do not address. For those areas, the scope of the submissions may include but is not limited to the following general topics:

    Boundary extraction from 2D/3D shapes

    Geometric deep learning on 3D and higher dimensions

    Generative methods for parametric representations

    Novel shape descriptors and embedding for geometric deep learning

    Deep learning on non-Euclidean geometries

    Transformation invariant shape abstractions

    Shape abstraction in different domains

    Synthetic data generation for data augmentation in geometric deep learning

    Comparison of shape representations for efficient deep learning

    Novel kernels and architectures specifically for 3D generative models

    Eigen-spectra analysis and graph-based approaches for 3D data

    Applications of geometric deep learning in different domains

    Learning-based estimation of shape differential quantities

    Detection of geometric feature lines from 3D data, including 3D point clouds and depth images

    Geometric shape segmentation, including patch decomposition and sharp lines detection

The CMT site for paper submissions is https://cmt3.research.microsoft.com/DLGC2021 . Each submitted paper must be 4-8 pages excluding references. Please refer to the ICCV author submission guidelines for instructions at http://iccv2021.thecvf.com/node/4#submission-guidelines. The review process will be double-blind but the papers will be linked to any associated challenge submissions. Selected papers will be published in IEEE ICCVW proceedings, visible in IEEE Xplore and on the CVF Website.

*************** Awards ***************

The winning submission in each seven track will receive a prize (either cash or equipment) provided by the workshop sponsors. The top submissions in each category with accepted papers in the workshop will be chosen as finalists and will be invited to present their research in the spotlight session.

*************** Important dates ***************

    Challenges Launch for Submissions: May, 07, 2021

    Second Phase for Submissions: July, 22, 2021

    Challenges Close for Submissions: August 1, 2021

    Abstract Submission Deadline: July 26, 2021

    Paper Submission Deadline: August 1, 2021

    Acceptance Notification: August 11, 2021

    Camera Ready Due: August 17, 2021

    Workshop (full day): October 11, 2021

***************Organizers***************

    Ilke Demir, Sr. Staff Research Scientist, Intel Corporation

    Alexey Artemov, Research Scientist, Skolkovo Institute of Science and Technology

    Dena Bazazian, Senior Research Associate, University of Bristol

    Bernhard Egger, Postdoctoral Researcher, MIT

    Géraldine Morin, Professor, University of Toulouse

    Kathryn Leonard, Professor of Computer Science, Occidental College

    Evgeny Burnaev, Associate Professor, Skolkovo Institute of Science and Technology

    Adarsh Krishnamurthy, Associate Professor, Iowa State University

    Daniele Panozzo, Assistant Professor, Courant Institute of Mathematical Sciences, New York University

    Albert Matveev, Ph.D. student, Skolkovo Institute of Science and Technology

    Denis Zorin, Professor of Computer Science and Mathematics, Chair of Computer Science Department Courant Institute of Mathematical Sciences, New York University

    Rana Hanocka, Ph.D. student, Tel Aviv University