March 23rd, 2023 
Daniela Lopez de Luise Call for Participants – Robotic Vision Scene Understanding Challenge – CVPR 2023
===================================================================
We are happy to announce the latest iteration of the Robotic Vision Scene Understanding (RVSU) challenge, running as part of the 4th Embodied AI Workshop (EAI4) at the 2023 Conference on Computer Vision and Pattern Recognition (CVPR 2023) has recently begun!
Evaluation servers are now open on EvalAI, accepting results from the next generation of active scene understanding systems.
Top two performing teams to receive GPU prizes and have their work highlighted during the workshop.
Register your interest: https://docs.google.com/forms/d/e/1FAIpQLSeNY2r-cEYC2cIlat4VunjmosrFV-5HYck9mY_C26w6Tjwl2w/viewform?usp=sf_link
Details of the challenge including important links, prize information and current known dates are outlined below.
Overview
========
The CVPR2023 Embodied AI Workshop of our Robotic Vision Scene Understanding Challenge evaluates how well a robotic vision system can understand the semantic and geometric aspects of environments. The challenge is performed in simulation and consists of two distinct tasks: Object-based Semantic SLAM, and Scene Change Detection.
- Semantic SLAM: Participants use a robot to traverse around environments, building up an object-based semantic map from the robot’s RGBD sensor observations and odomtry measurements.
- Scene change detection (SCD): Participants use a robot to traverse through an environment scene, building up a semantic understanding of the scene. Then the robot is moved to a new start position in the same environment, but with different conditions. Along with a possible change from day to night, the new scene has a number objects added and / or removed. Participants must produce an object-based semantic map describing the changes between the two scenes.
Each task has two difficulty levels where the agent is supplied with either ground-truth or noisy pose data.
The challenge is run using the BenchBot framework on top of NVIDIA's Isaac Simulator to render BenchBot Environments for Active Robotics (BEAR). Participants need to program robotic agents with an OpenAI Gym-style API to actively explore these environments and solve the given tasks.
Prizes
=====
1 RTX A6000 and up to 5 Jetson Nanos (1 for each member) for each of the top 2 winning teams.
Important Dates
=============
- March 15th – Challenge launch
- May – Submissions due
- June 19th – CVPR 2023 Embodied AI workshop
Important Links
=============
Challenge Website: https://nikosuenderhauf.github.io/roboticvisionchallenges/cvpr2023
Interest Registration Form: https://docs.google.com/forms/d/e/1FAIpQLSeNY2r-cEYC2cIlat4VunjmosrFV-5HYck9mY_C26w6Tjwl2w/viewform?usp=sf_link
BenchBot Webpage: https://github.com/qcr/benchbot
Embodied AI Workshop (2022 page): https://embodied-ai.org/
Challenge Submission Page: https://eval.ai/web/challenges/challenge-page/1614/overview
Contact Us
=========
Slack Channel: https://app.slack.com/client/T012ZHR4CQG
Twitter: @robVisChallenge – https://twitter.com/robVisChallenge
Organizer e-mail: david.hall@csiro.au
General contact e-mail: contact@roboticvisionchallenge.org
Extra Links
=========
YouTube challenge overview (2019): https://youtu.be/jQPkV29KFvI
Dr David Hall (he/him)
Research Scientist
Posted in