I do not find images with bikes, riders, motos or trains

[RobertoAlcoverCouso]

I have downloaded the dataset and i find 3 missing classes: riders, motos and trains. Furthermore, the class bike is only in around 30 images and those 30 images contain around 40 pixels of class bike.
Is there a mistake with the published dataset?

[giannifranchi]

Hi Roberto, I wanted to bring to your attention that the training set does not include the classes riders, motos, and trains. These three classes represent out-of-distribution (OOD) classes, and the challenge aims to evaluate how well your DNN performs when it encounters these unseen classes. Furthermore, there are additional peculiar OOD classes such as cow, rock, deer, and so on. I hope this clarifies the objective. We strive to adhere to normal statistics, and it is indeed challenging to handle the bike class.

[farhadi76m]

Hi Roberto, I wanted to bring to your attention that the training set does not include the classes riders, motos, and trains. These three classes represent out-of-distribution (OOD) classes, and the challenge aims to evaluate how well your DNN performs when it encounters these unseen classes. Furthermore, there are additional peculiar OOD classes such as cow, rock, deer, and so on. I hope this clarifies the objective. We strive to adhere to normal statistics, and it is indeed challenging to handle the bike class.

Hello, I have to say, we need a grand truth for ood data, since I don't know how to evaluate them .