Hey, looks like there are just 4 different boast, with the same boats in train and test sets. Will there be an additional test set with different boats, or we should just overfit to the 4 boats we have?
Hi @lopuhin, the goal of the challenge is to create models which will be useful in new contexts beyond the data available currently. As always, best practices are encouraged to ensure a model will generalize, and assumptions should be documented for the organizations who will use the models. There will not be a release of additional data during the competition.
Thanks, and have fun!
High score can be achieved by sticking to these boats (e.g. By classifiying each boat). Will that solution be rejected on evaluation step?