All videos (invited talks, new competition, accepted papers) are now accessible on the ECCV virtual platform (registration required): https://workshopsandtutorials.eccv2020.eu/papers/subject/perception-for-autonomous-driving/
Autonomous Driving (AD) has the potential to revolutionize mobility and bring lasting benefits to society. It is thus at the forefront of AI research and has attracted the attention of both academia and industry. As an example, half of the exhibitions at CVPR 2019 are related to AD. From a Computer Vision perspective, the most relevant task in AD is Perception, i.e. understanding the world around the car. After discussions with both academic researchers and industrial practitioners, we feel that the temporal and multi-modal aspects of perception have been overlooked. Robust tracking and more importantly prediction of movement, for both vehicles and pedestrians, are critical for AD. This issue is particularly acute in dense urban environments, which are heterogeneous multi-agent systems consisting of diverse traffic participants with a great variety of shapes, dynamics, behaviors, and intents.
Workshop highlight & schedule
11 invited talks + 2 Panel discussions and Q&A
1 new competition in motion prediction will be announced
20 accepted paper + 2 Live Q&A sessions
All videos (invited talks, new competition, accepted papers) are accessible on the ECCV virtual platform now (registration required): https://workshopsandtutorials.eccv2020.eu/papers/subject/perception-for-autonomous-driving/. We recommend that you watch them and come to discuss with the speakers and authors!
Sunday August 23rd
Session 1: 09:45 - 12:00 UTC+1
09:45 - 09:48 UTC+1 Open remark
09:48 - 10:00 UTC+1 Competition announcement: Motion forecasting for self-driving
10:00 - 11:00 UTC+1 Panel discussion and Q&A with invited speakers (Deva Ramanan, Antonio Lopez, Alexandre Alahi, Wolfram Burgard)
11:00 - 12:00 UTC+1 Paper Q&A
Session 2: 17:45 - 20:00 UTC+1
17:45 - 17:48 UTC+1 Open remark
17:48 - 18:00 UTC+1 Competition announcement: Motion forecasting for self-driving
18:00 - 19:00 UTC+1 Panel discussion and Q&A with invited speakers (Alyssa Pierson, Raquel Urtasun, Urs Muller, Anelia Angelova, Andreas Geiger, Peter Kontschieder, Antonio Lopez)
19:00 - 20:00 UTC+1 Paper Q&A
University of Toronto, Uber ATG
We will be starting a new competition at the workshop focused on motion prediction:
Abstract: Motion Prediction (ability to predict the future behaviour of detected objects around AV) is quickly becoming the next frontier of AV research. In this session, we will go over the new motion prediction competition starting this week at ECCV and sponsored by Lyft Level 5 with $30k in prizes. Summary: Task: Predict future motion of vehicles around AV Dataset: Lyft Prediction Dataset 2020 Starting: 24th August (tomorrow) Duration: 3 months Prizes: $30k Run at Kaggle Sponsored and organised by Lyft Level 5
Task: Motion forecasting for self-driving
Dataset: Lyft prediction dataset 2020
Duration: 3 months
Call for Contributions
Interested researchers from both, academia or industry are invited to submit non-anonymous extended abstracts.
Page limit: 2-4 pages of main+references, and 0+ pages of supplementary, as a single PDF
Submission Website: https://cmt3.research.microsoft.com/PAD2020
LaTeX Template: https://eccv2020.eu/wp-content/uploads/2020/01/eccv2020kit-1.zip (for submission only)
Camera-ready submission information:
The camera-ready submission will include: (1) main paper, (2) (optional) supplementary material, (3) 3-min video, and (4) the slides for the 3-min video.
Main paper: at most "5" pages including references, in a PDF named "PAD-XX.pdf", where XX stands for the paper ID (e.g., 01, 11).
Supplementary material: You may upload the supplementary material in a PDF named "PAD-XX-suppl.pdf". There is no limit on the number of pages.
3-min video: mp4 format
Slides: in a PDF named "PAD-XX-slides.pdf"
Camera-ready versions that do not follow the page limit may not be posted on our website and will not be presented in the workshop.
Accepted papers will be presented in a 3-min recorded video and a live Q&A format.
Accepted papers will be posted on the website, but there will not be archival proceedings.
Submit by: 23rd July 2020 AoE
Notification: 4th August 2020
Camera ready: 16th August 2020
Workshop: 23rd August 2020
A special thanks to the program committee for making this possible!
Li Erran Li