Dear Organizers,
I was going through the Challenge website and the Codalab description. The challenge statement is not very clear.
On the codalab website, it is written "The objective is to evaluate self-supervised monocular depth estimation on the challenging SYNS-Patches dataset. ".
1. Does this means we have to evaluate the "EXISTING METHODS" and find the answers to the three questions mentioned here (https://jspenmar.github.io/MDEC/#challenge). ?
2. Or we have to propose a "NEW METHOD" whose performance should be at least Garg baseline [13.7211] (by jspenmar) and analyze in the context of the three questions mentioned here (https://jspenmar.github.io/MDEC/#challenge)?.
Please clarify.
Thanks,
Lokender
Dear Lokender,
Thank you for your interest in the competition. The three questions listed in the workshop website describe our motivation behind organizing the workshop and challenge, you do not need to answer them.
For this challenge, you need to submit a method developed by you. This could be a preexisting method (as long as it was developed by you) or a novel method designed specifically for this challenge.
Our evaluation script will compute the metrics required on the SYNS-Patches dataset, and submissions will be ranked based on the F-Score performance.
I hope this helps to clarify things.
Kind regards,
Jaime