Add missing entry of 'isthing' for coco panoptic in customize dataset

[FranklynJey]

Motivation

I created a customized dataset for coco panoptic segmentation by following mmdetection/docs/en/advanced_guides/customize_dataset.md.

In the description, we can find

categories': [  # including both foreground categories and background categories
    {'id': 0, 'name': 'person'},
    ...

A dataset created like this will cause errors, as the attribute 'isthing' is required. I first realized it in browse_dataset.py which breaks using the panoptic coco dataset format according to customized_dataset.md.

See

This holds even if metainfo is defined as

metainfo = dict(
        classes = ('class1','class2',... ),
        thing_classes = ('class1',... ),
        stuff_classes = ('class2',...)
        )

and is given to train_dataloader etc.

Modification

I would recommend changing the line

categories': [  # including both foreground categories and background categories
    {'id': 0, 'name': 'person'},

to

categories': [  # including both foreground categories and background categories
    {'id': 0, 'name': 'person', 'isthing': 1},

so that it matches the original coco panoptic description and prevents errors in mmdetection.

PS

Thank you guys so much for your awesome work in mmdetection!

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
3. If the modification has potential influence on downstream projects, this PR should be tested with downstream projects, like MMDet or MMPreTrain.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[CLAassistant]

All committers have signed the CLA.

[FranklynJey]

@jbwang1997 did you have time to look on it ? :)