Call for Papers
The workshop seeks for contributions describing new methodologies allowing to estimate objects’ locations and pose in the 3D world as well as novel benchmarks and resources to automatically generate and annotate large scale datasets. Contributions discussing novel evaluation protocols and metrics to fairly compare 3D object detectors are also welcome.
In order to make broader stage for relevant contributions to the field, the workshop will also accept main conference submissions. All the accepted submissions will be presented in our poster session. The best submissions will also be presented in an oral session and will be eligible for the best paper prizes!
Topics
The suggested list of topics is (but is not limited to):
3D object detection methods from images in a monocular setting, stereo or other calibrated multi-camera settings, or from multiple image frames or video streams
Exploration of novel representations, e.g. like Pseudo-LiDAR, for image-based 3D object detection
Exploitation of semi-supervised or self-supervised techniques in the context of image-based 3D object detection
Domain adaptation or transfer learning with image-based 3D object detectors
Exploring adversarial attacks in the context of image-based 3D object detection
Advances on depth estimation for the purpose of improving 3D Object Detection pipelines
Novel metrics to benchmark 3D object detectors as well as critical analyses of the existing metrics and evaluation protocols
Novel benchmark datasets for image-based 3D object detection methods, or critical analyses of existing benchmarks and methods
Novel strategies for investigating the limits of existing monocular 3D object detection models
Applications of image-based 3D object detections (robotics, autonomous driving, Augmented reality, etc.)
Important dates
The main deadlines are:
Submission:
July 25th 11:59PM Pacific TimeNotification:
August 12th11:59PM Pacific TimeCamera ready:
August 16th 11:59PM Pacific Time
Submission Guidelines
The guidelines for the submission are:
Template: all the submitted manuscripts must follow the ICCV21 template
Page limit: 4-8 pages excluding references
Submission: papers must be submitted to CMT at https://cmt3.research.microsoft.com/3DODI2021
Publication: accepted papers will appear in the main conference proceedings and on the workshop page
Presentation: accepted papers will be presented as Oral or Poster presentations during the workshop
Policy: papers will be peer reviewed under double-blind policy. Papers will be rejected without review if they contain more than 8 pages (w/o references) or violate the double-blind policy
Main conference submissions: we accept ICCV21 main conference submissions. These submissions will not be eligible for prizes. Authors must declare if their paper is part of the main conference at the time of submission to the workshop
Prizes
In order to solicit high-quality submissions we will have two best paper prizes:
Best paper: 2K $
Runner-up: 1K $
Note: prizes will be given to 3DODI-only submissions, main-conference or double submissions are not eligible
Contacts: for any information email to 3dodi.workshop@gmail.com