Image Matching: Local Features & Beyond
CVPR 2019 Workshop
Sunday 16th June 2019
Hyatt Regency A
- 08:45-09:00 Welcome
- 09:00-09:30 Amir Zamir (Stanford/UC Berkeley), Collection of Large-scale Densely-labeled 3D Data from the Real World Without a Single Click
- 09:30-10:15 Jiri Matas (CTU Prague), On the Art of Establishing Correspondence (slides)
- 10:15-11:00 Coffee Break & Poster Session
- 11:15-12:00 Torsten Sattler (Chalmers University of Technology), In Defense of Local Features for Visual Localization
- 12:00-12:15 IMW 2019 Challenge
- 12:15-12:30 Zixin Luo (HKUST), Winner of the Phototourism Challenge (slides)
- 12:30-12:45 Awards and Closing
The workshop papers are available on CVF: link.
Traditional, keypoint-based formulations for image matching are still
very competitive on tasks such as Structure from Motion (SfM), despite
recent efforts on tackling pose estimation with dense networks. In
practice, many state-of-the-art pipelines still rely on methods that
stood the test of time, such as SIFT or RANSAC.
In this workshop, we aim to encourage novel strategies for image
matching that deviate from and advance traditional formulations, with
a focus on large-scale, wide-baseline matching for 3D reconstruction
or pose estimation. This can be achieved by applying new technologies
to sparse feature matching, or doing away with keypoints and
The workshop topics include (but are not limited to):
- Reformulating keypoint extraction and matching pipelines with deep
- Applying geometric constraints into the training of (sparse or
dense) deep networks.
- Leveraging additional cues such as semantics.
- Developing adversarial methods to deal with conditions where current
methods fail (weather changes, day versus night, etc.).
- Exploring attention mechanisms to match salient image regions.
- Integrating differentiable components into 3D reconstruction
- Matching across different data modalities such as aerial versus
University of Bordeaux
University of Victoria
The workshop will be held on Sunday morning, June 16, 2019.
Please contact us on firstname.lastname@example.org.
Chalmers University of Technology
Stanford & Berkeley
Czech Technical University
- June 2: Uploaded the final set of results for the PT challenge (apologies for the delay, we had a few last-minute submissions). Congratulations to Dawei Sun, Zixin Luo, and Jiahui Zhang for winning both tracks with their ContextDesc descriptors and a custom, deep network-based matcher! See you at the conference.
- May 20: Removed D2-Net, as its training data overlaps with our test data.
- May 16: Removed a few bugged results (SIFT below 8k points).
- May 15: Added Brisk+SSS to the Phototourism leaderboards, fixed a bug in the aggregation of results for the MVS task.
- May 10: Added ELF, ContextDesc and SIFT-AID/NN to the Phototourism leaderboards.
- May 8: Added more entries to the Phototourism leaderboards.
- May 7: Added D2-Net to the Phototourism leaderboards.
- May 6: Added DELF to the Phototourism leaderboards.
- April 27: Paper submission is closed. Updated the leaderboards with a few more entries.
- April 25: Updated the leaderboards for the Phototourism dataset. More entries coming soon.
- April 24: Updated submission instructions for the PhotoTourism dataset, please follow this format in future submissions.
- April 23: Uploaded data and submission instructions for the SILDa image matching dataset.
- April 11: Uploaded training data for the Phototourism dataset.
- March 27: Due to demand and to the proximity with the ICCV deadline, we are
moving the paper submission deadline to April 26.
- February 28: Uploaded test data and submission instructions for the
- February 22: Moving the paper submission deadline to March 24.
- February 18: Confirmed Prof. Jiri Matas as a keynote speaker.
- February 15: Uploaded new draft with a CFP and some challenge entries. More to
- Paper submission deadline: March 24, 2019 April 26, 2019
- Notification to authors: April 5, 2019 May 7, 2019
- Camera-ready deadline: April 10, 2019 May 14, 2019
- Challenge and extended abstracts deadline: May 20, 2019
(Deadlines are 11:59PM in Pacific Time.)
The workshop features a challenge with multiple datasets and tasks.
Please refer to the Call for Papers.
The prizes for the workshop challenge are sponsored by Google and Scape.