[Features] Add support for Kitti semantic segmentation dataset

[AkideLiu]

Describe the feature

Add support for KITTI semantic segmentation dataset

Motivation

Kitti semantic segmentation dataset is a lightweight dataset for semantic segmentation which shares the same label policy as cityscapes. It's an excellent starting point for segmentation and employs the weights pre-trained on cityscapes to perform transfer-learning, do you consider to support this dataset.

Related resources

http://www.cvlibs.net/datasets/kitti/eval_semseg.php?benchmark=semantics2015

Additional context

Happy to create a PR.

[MeowZheng]

Thanks in advance for your contribution to mmseg, and looking forward to your nice PR.

[AkideLiu]

Thanks in advance for your contribution to mmseg, and looking forward to your nice PR.

Thanks for your quick reply, I have the following questions looking forward to your reply.

As this dataset has 200 training images with the fine label, however, the test set is not provided, do I need to implement the train_test_split for this dataset. In my current using scenario, the training set has been split from the original training set with 150 images and 50 images for both testing and validation datasets.

if the train_test_split is required, could you please provide a guideline for the split ratio and reference code for insight?