Dear Maurice,
Nice that you have a team of undergraduates that want to participate in the Virtual Robots competition.
As co-author of some of the papers I can give you some background. Many of the papers can be found here:
https://rescuesim.robocup.org/research/publications/virtual-robot-competitio...
The one that is not on the list, but often used as starting point of recognition of victims, is Helen Flynn's Msc-thesis
https://www.researchgate.net/publication/224773218_Machine_Learning_Applied_...
In 2009 the victims still had a RFID tag, which could be used to supervise the learning. This tag was removed in later versions, and the dataset of the images used for training was never published.
The same is truth for many other cases. The worlds created by the Technical Committee are published, but the teams have not published (as far as I known) annotated datasets, although that would be very useful.
In that sense, we can learn a lot of the other RoboCup leagues, as for instance the SPL:
https://spl.robocup.org/datasets/
Regards,
Arnoud Visser https://staff.fnwi.uva.nl/a.visser/research/roboresc/
Van: robocup-rescue-v r**********************s@lists.robocup.org Namens Maurice Pagnucco Verzonden: woensdag 4 augustus 2021 13:20 Aan: r**************v@lists.robocup.org Onderwerp: [robocup-rescue-v] Virtual Robots Competition - Requesting Assistance
Dear Virtual Robot Competitors,
I'm from the School of Computer Science and Engineering at the University of New South Wales in Sydney, Australia. We have a long history of competing in RoboCup, particular the Standard Platform League in soccer, RoboCup Rescue and RoboCup@Home.
We're currently in the process of starting a new team to hopefully compete in the RoboCup Rescue Virtual Robots Competition next year (2022) and have spent a few months familiarising ourselves with the competition. However, we could do with some help and guidance.
I have put together a team of undergraduate students from my university who are interested in participating in the competition The students have spent the better half of this year investigating more about the competition and trying to learn the technical skills needed including ROS, Gazebo, etc. A group of students from the team are focusing on the object recognition part of the competition and experimenting with technologies such as YOLO and Detectron2 to detect victims in the maps. We went through papers from the past few year submitted by previous participating teams and noticed that most of them use a dataset of victims to train their image recognition models. However, we could not find the source of the datasets anywhere in the papers, internet or the RoboCup website. We did find some soccer datasets but none is specific for the object recognition task.
Would someone be able to let me know if there are any such technical datasets/resources that we can access to assist with our participation in the competition next year?
I also noticed that training workshops were held for some of the previous iterations of this competition. Is there any such training workshop(s) planned for next year or can we access the material from one of the previous workshops?
Lastly, if anyone from the community would be interested in presenting to our group over videoconference at some point, we would be very grateful.
Thanks and best wishes to all, Morri. __ Professor Maurice Pagnucco Deputy Dean (Education), UNSW Engineering Deputy Director, Creative Robotics Laboratory PLuS Alliance Fellow Professorial Fellow, UNSW iCinema Centre for Interactive Cinema Research UNSW Sydney
Dean's Unit, Faculty of Engineering Level 6, Building K17 UNSW SYDNEY 2052
E: m***i@cse.unsw.edu.aumailto:m***i@cse.unsw.edu.au, m********o@unsw.edu.aumailto:m********o@unsw.edu.au W: unsw.edu.auhttps://eur04.safelinks.protection.outlook.com/?url=http%3A%2F%2Funsw.edu.au%2F&data=04%7C01%7Ca.visser%40uva.nl%7C3b4d29a0266141cd8d0908d9573a27bd%7Ca0f1cacd618c4403b94576fb3d6874e5%7C1%7C1%7C637636729562851787%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=8x%2BQNjlAg2PLApGwrTeFTnHyL1RohGvh00G18wILc%2BQ%3D&reserved=0 T: +61 (0)2 9385 5000 Executive Assistant: Rachel Lewis E: r**********s@unsw.edu.aumailto:r**********s@unsw.edu.au
[cid:i**********g@01D79293.8F9A4960]
CRICOS Provider Code 00098G
