数据集划分

[wxmchgithub]

为什么自定义数据集只分了train和val呢，不需要test集吗，有大佬解答一下吗，感谢！

[mm-assistant]

We recommend using English or English & Chinese for issues so that we could have broader discussion.

[wxmchgithub]

Why is the custom data set only divided into train and val? Do not need the test set？

[MeowZheng]

Could you tell me which documentation includes this ?

custom data set only divided into train and val

Some datasets don't release test sub-dataset or don't release the ground truth of test sub-dataset, and we cannot evaluate model locally without gt of test sub-dataset, so we ignore test sub-dataset in cfg files.

However, we will never prevent you to split your dataset into train val and test sub-datasets,

[wxmchgithub]

Could you tell me which documentation includes this ?
docs/en/dataset_prepare.md

custom data set only divided into train and val

Some datasets don't release test sub-dataset or don't release the ground truth of test sub-dataset, and we cannot evaluate model locally without gt of test sub-dataset, so we ignore test sub-dataset in cfg files.
The data organization of these examples is divided into train and val, without test. I want to split train, val, test. it's ok?
However, we will never prevent you to split your dataset into train val and test sub-datasets,