[Feature] add bdd100K datasets #3158
Conversation
CastleDream
changed the title
CastleDream
changed the title
CastleDream
changed the title
mmseg/datasets/bdd100k.py
Outdated
| classes=('road', 'sidewalk', 'building', 'wall', 'fence', 'pole', | ||
| 'traffic light', 'traffic sign', 'vegetation', 'terrain', | ||
| 'sky', 'person', 'rider', 'car', 'truck', 'bus', 'train', | ||
| 'motorcycle', 'bicycle', 'Unlabeled'), | ||
| palette=[[128, 64, 128], [244, 35, 232], [70, 70, 70], [102, 102, 156], | ||
| [190, 153, 153], [153, 153, 153], [250, 170, | ||
| 30], [220, 220, 0], | ||
| [107, 142, 35], [152, 251, 152], [70, 130, 180], | ||
| [220, 20, 60], [255, 0, 0], [0, 0, 142], [0, 0, 70], | ||
| [0, 60, 100], [0, 80, 100], [0, 0, 230], [119, 11, 32], | ||
| [0, 0, 0]]) |
There was a problem hiding this comment.
Could we remove the Unlabeled class?
There was a problem hiding this comment.
Yes，we should keep consistent with Cityscapes. In [bdd100k/label.py]
(https://github.com/bdd100k/bdd100k/blob/master/bdd100k/label/label.py#L42C47-L42C47) mentions:
# Our extended list of label types. Our train id is compatible with Cityscapes
labels = [
# name id trainId category catId
# hasInstances ignoreInEval color
Label("unlabeled", 0, 255, "void", 0, False, True, (0, 0, 0)),but i am confused that why some dataset.py, for example in datasets/mapillary.py, including Unlabeled, and most other dataset.py, didn't include.
and in the openMMLabCamp classes video, the teacher also add Unlabeled when import custom dataset..
There was a problem hiding this comment.
This might depend on the requirements at the time of creating the dataset, and if there is a need to accurately recognize the background class, then it is necessary to include this class.
There was a problem hiding this comment.
Fine, thanks for your apply!
xiexinch
changed the title
## Motivation Integrate [BDD100K](https://paperswithcode.com/dataset/bdd100k) dataset. It shares the same classes as Cityscapes, and it's commonly used for evaluating segmentation/detection tasks in driving scenes, such as in [RobustNet](https://arxiv.org/abs/2103.15597), [WildNet](https://github.com/suhyeonlee/WildNet). Enhancement for Add BDD100K Dataset open-mmlab#2808 --------- Co-authored-by: xiexinch <[email protected]>
Motivation
Integrate BDD100K dataset. It shares the same classes as Cityscapes, and it's commonly used for evaluating segmentation/detection tasks in driving scenes, such as in RobustNet, WildNet.
Enhancement for Add BDD100K Dataset #2808