[Feature] Add MultiImageMixDataset (#1105)

* Fix typo in usage example

* original MultiImageMixDataset code in mmdet

* Add MultiImageMixDataset unittests in test_dataset_wrapper

* fix lint error

* fix value name ann_file to ann_dir

* modify retrieve_data_cfg (#1)

* remove dynamic_scale & add palette

* modify retrieve_data_cfg method

* modify retrieve_data_cfg func

* fix error

* improve the unittests coverage

* fix unittests error

* Dataset (#2)

* add cfg-options

* Add unittest in test_build_dataset

* add blank line

* add blank line

* add a blank line

Co-authored-by: Miao Zheng <76149310+MeowZheng@users.noreply.github.com>

Co-authored-by: Younghoon-Lee <72462227+Younghoon-Lee@users.noreply.github.com>
Co-authored-by: MeowZheng <meowzheng@outlook.com>
Co-authored-by: Miao Zheng <76149310+MeowZheng@users.noreply.github.com>

mmseg/datasets/__init__.py

@@ -6,7 +6,8 @@ from .cityscapes import CityscapesDataset
 from .coco_stuff import COCOStuffDataset
 from .custom import CustomDataset
 from .dark_zurich import DarkZurichDataset
-from .dataset_wrappers import ConcatDataset, RepeatDataset
+from .dataset_wrappers import (ConcatDataset, MultiImageMixDataset,
+                               RepeatDataset)
 from .drive import DRIVEDataset
 from .hrf import HRFDataset
 from .loveda import LoveDADataset
@@ -21,5 +22,5 @@ __all__ = [
     'PascalVOCDataset', 'ADE20KDataset', 'PascalContextDataset',
     'PascalContextDataset59', 'ChaseDB1Dataset', 'DRIVEDataset', 'HRFDataset',
     'STAREDataset', 'DarkZurichDataset', 'NightDrivingDataset',
-    'COCOStuffDataset', 'LoveDADataset'
+    'COCOStuffDataset', 'LoveDADataset', 'MultiImageMixDataset'
 ]

mmseg/datasets/builder.py

@@ -64,12 +64,18 @@ def _concat_dataset(cfg, default_args=None):
 
 def build_dataset(cfg, default_args=None):
     """Build datasets."""
-    from .dataset_wrappers import ConcatDataset, RepeatDataset
+    from .dataset_wrappers import (ConcatDataset, RepeatDataset,
+                                   MultiImageMixDataset)
     if isinstance(cfg, (list, tuple)):
         dataset = ConcatDataset([build_dataset(c, default_args) for c in cfg])
     elif cfg['type'] == 'RepeatDataset':
         dataset = RepeatDataset(
             build_dataset(cfg['dataset'], default_args), cfg['times'])
+    elif cfg['type'] == 'MultiImageMixDataset':
+        cp_cfg = copy.deepcopy(cfg)
+        cp_cfg['dataset'] = build_dataset(cp_cfg['dataset'])
+        cp_cfg.pop('type')
+        dataset = MultiImageMixDataset(**cp_cfg)
     elif isinstance(cfg.get('img_dir'), (list, tuple)) or isinstance(
             cfg.get('split', None), (list, tuple)):
         dataset = _concat_dataset(cfg, default_args)

mmseg/datasets/dataset_wrappers.py

@@ -1,13 +1,15 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 import bisect
+import collections
+import copy
 from itertools import chain
 
 import mmcv
 import numpy as np
-from mmcv.utils import print_log
+from mmcv.utils import build_from_cfg, print_log
 from torch.utils.data.dataset import ConcatDataset as _ConcatDataset
 
-from .builder import DATASETS
+from .builder import DATASETS, PIPELINES
 from .cityscapes import CityscapesDataset
 
 
@@ -188,3 +190,88 @@ class RepeatDataset(object):
     def __len__(self):
         """The length is multiplied by ``times``"""
         return self.times * self._ori_len
+
+
+@DATASETS.register_module()
+class MultiImageMixDataset:
+    """A wrapper of multiple images mixed dataset.
+
+    Suitable for training on multiple images mixed data augmentation like
+    mosaic and mixup. For the augmentation pipeline of mixed image data,
+    the `get_indexes` method needs to be provided to obtain the image
+    indexes, and you can set `skip_flags` to change the pipeline running
+    process.
+
+
+    Args:
+        dataset (:obj:`CustomDataset`): The dataset to be mixed.
+        pipeline (Sequence[dict]): Sequence of transform object or
+            config dict to be composed.
+        skip_type_keys (list[str], optional): Sequence of type string to
+            be skip pipeline. Default to None.
+    """
+
+    def __init__(self, dataset, pipeline, skip_type_keys=None):
+        assert isinstance(pipeline, collections.abc.Sequence)
+        if skip_type_keys is not None:
+            assert all([
+                isinstance(skip_type_key, str)
+                for skip_type_key in skip_type_keys
+            ])
+        self._skip_type_keys = skip_type_keys
+
+        self.pipeline = []
+        self.pipeline_types = []
+        for transform in pipeline:
+            if isinstance(transform, dict):
+                self.pipeline_types.append(transform['type'])
+                transform = build_from_cfg(transform, PIPELINES)
+                self.pipeline.append(transform)
+            else:
+                raise TypeError('pipeline must be a dict')
+
+        self.dataset = dataset
+        self.CLASSES = dataset.CLASSES
+        self.PALETTE = dataset.PALETTE
+        self.num_samples = len(dataset)
+
+    def __len__(self):
+        return self.num_samples
+
+    def __getitem__(self, idx):
+        results = copy.deepcopy(self.dataset[idx])
+        for (transform, transform_type) in zip(self.pipeline,
+                                               self.pipeline_types):
+            if self._skip_type_keys is not None and \
+                    transform_type in self._skip_type_keys:
+                continue
+
+            if hasattr(transform, 'get_indexes'):
+                indexes = transform.get_indexes(self.dataset)
+                if not isinstance(indexes, collections.abc.Sequence):
+                    indexes = [indexes]
+                mix_results = [
+                    copy.deepcopy(self.dataset[index]) for index in indexes
+                ]
+                results['mix_results'] = mix_results
+
+            results = transform(results)
+
+            if 'mix_results' in results:
+                results.pop('mix_results')
+
+        return results
+
+    def update_skip_type_keys(self, skip_type_keys):
+        """Update skip_type_keys.
+
+        It is called by an external hook.
+
+        Args:
+            skip_type_keys (list[str], optional): Sequence of type
+                string to be skip pipeline.
+        """
+        assert all([
+            isinstance(skip_type_key, str) for skip_type_key in skip_type_keys
+        ])
+        self._skip_type_keys = skip_type_keys

tests/test_data/test_dataset.py

@@ -14,7 +14,8 @@ from PIL import Image
 from mmseg.core.evaluation import get_classes, get_palette
 from mmseg.datasets import (DATASETS, ADE20KDataset, CityscapesDataset,
                             ConcatDataset, CustomDataset, LoveDADataset,
-                            PascalVOCDataset, RepeatDataset, build_dataset)
+                            MultiImageMixDataset, PascalVOCDataset,
+                            RepeatDataset, build_dataset)
 
 
 def test_classes():
@@ -95,6 +96,66 @@ def test_dataset_wrapper():
     assert repeat_dataset[27] == 7
     assert len(repeat_dataset) == 10 * len(dataset_a)
 
+    img_scale = (60, 60)
+    pipeline = [
+        # dict(type='Mosaic', img_scale=img_scale, pad_val=255),
+        # need to merge mosaic
+        dict(type='RandomFlip', prob=0.5),
+        dict(type='Resize', img_scale=img_scale, keep_ratio=False),
+    ]
+
+    CustomDataset.load_annotations = MagicMock()
+    results = []
+    for _ in range(2):
+        height = np.random.randint(10, 30)
+        weight = np.random.randint(10, 30)
+        img = np.ones((height, weight, 3))
+        gt_semantic_seg = np.random.randint(5, size=(height, weight))
+        results.append(dict(gt_semantic_seg=gt_semantic_seg, img=img))
+
+    classes = ['0', '1', '2', '3', '4']
+    palette = [(0, 0, 0), (1, 1, 1), (2, 2, 2), (3, 3, 3), (4, 4, 4)]
+    CustomDataset.__getitem__ = MagicMock(side_effect=lambda idx: results[idx])
+    dataset_a = CustomDataset(
+        img_dir=MagicMock(),
+        pipeline=[],
+        test_mode=True,
+        classes=classes,
+        palette=palette)
+    len_a = 2
+    cat_ids_list_a = [
+        np.random.randint(0, 80, num).tolist()
+        for num in np.random.randint(1, 20, len_a)
+    ]
+    dataset_a.data_infos = MagicMock()
+    dataset_a.data_infos.__len__.return_value = len_a
+    dataset_a.get_cat_ids = MagicMock(
+        side_effect=lambda idx: cat_ids_list_a[idx])
+
+    multi_image_mix_dataset = MultiImageMixDataset(dataset_a, pipeline)
+    assert len(multi_image_mix_dataset) == len(dataset_a)
+
+    for idx in range(len_a):
+        results_ = multi_image_mix_dataset[idx]
+
+    # test skip_type_keys
+    multi_image_mix_dataset = MultiImageMixDataset(
+        dataset_a, pipeline, skip_type_keys=('RandomFlip'))
+    for idx in range(len_a):
+        results_ = multi_image_mix_dataset[idx]
+        assert results_['img'].shape == (img_scale[0], img_scale[1], 3)
+
+    skip_type_keys = ('RandomFlip', 'Resize')
+    multi_image_mix_dataset.update_skip_type_keys(skip_type_keys)
+    for idx in range(len_a):
+        results_ = multi_image_mix_dataset[idx]
+        assert results_['img'].shape[:2] != img_scale
+
+    # test pipeline
+    with pytest.raises(TypeError):
+        pipeline = [['Resize']]
+        multi_image_mix_dataset = MultiImageMixDataset(dataset_a, pipeline)
+
 
 def test_custom_dataset():
     img_norm_cfg = dict(

tests/test_data/test_dataset_builder.py

@@ -6,8 +6,8 @@ import pytest
 from torch.utils.data import (DistributedSampler, RandomSampler,
                               SequentialSampler)
 
-from mmseg.datasets import (DATASETS, ConcatDataset, build_dataloader,
+from mmseg.datasets import (DATASETS, ConcatDataset, MultiImageMixDataset,
-                            build_dataset)
+                            build_dataloader, build_dataset)
 
 
 @DATASETS.register_module()
@@ -48,6 +48,11 @@ def test_build_dataset():
     assert isinstance(dataset, ConcatDataset)
     assert len(dataset) == 10
 
+    cfg = dict(type='MultiImageMixDataset', dataset=cfg, pipeline=[])
+    dataset = build_dataset(cfg)
+    assert isinstance(dataset, MultiImageMixDataset)
+    assert len(dataset) == 10
+
     # with ann_dir, split
     cfg = dict(
         type='CustomDataset',

tools/browse_dataset.py

@@ -5,7 +5,7 @@ from pathlib import Path
 
 import mmcv
 import numpy as np
-from mmcv import Config
+from mmcv import Config, DictAction
 
 from mmseg.datasets.builder import build_dataset
 
@@ -42,6 +42,16 @@ def parse_args():
         type=float,
         default=0.5,
         help='the opacity of semantic map')
+    parser.add_argument(
+        '--cfg-options',
+        nargs='+',
+        action=DictAction,
+        help='override some settings in the used config, the key-value pair '
+        'in xxx=yyy format will be merged into config file. If the value to '
+        'be overwritten is a list, it should be like key="[a,b]" or key=a,b '
+        'It also allows nested list/tuple values, e.g. key="[(a,b),(c,d)]" '
+        'Note that the quotation marks are necessary and that no white space '
+        'is allowed.')
     args = parser.parse_args()
     return args
 
@@ -122,28 +132,32 @@ def _retrieve_data_cfg(_data_cfg, skip_type, show_origin):
         ]
 
 
-def retrieve_data_cfg(config_path, skip_type, show_origin=False):
+def retrieve_data_cfg(config_path, skip_type, cfg_options, show_origin=False):
     cfg = Config.fromfile(config_path)
+    if cfg_options is not None:
+        cfg.merge_from_dict(cfg_options)
     train_data_cfg = cfg.data.train
     if isinstance(train_data_cfg, list):
         for _data_cfg in train_data_cfg:
+            while 'dataset' in _data_cfg and _data_cfg[
+                    'type'] != 'MultiImageMixDataset':
+                _data_cfg = _data_cfg['dataset']
             if 'pipeline' in _data_cfg:
                 _retrieve_data_cfg(_data_cfg, skip_type, show_origin)
-            elif 'dataset' in _data_cfg:
-                _retrieve_data_cfg(_data_cfg['dataset'], skip_type,
-                                   show_origin)
             else:
                 raise ValueError
-    elif 'dataset' in train_data_cfg:
-        _retrieve_data_cfg(train_data_cfg['dataset'], skip_type, show_origin)
     else:
+        while 'dataset' in train_data_cfg and train_data_cfg[
+                'type'] != 'MultiImageMixDataset':
+            train_data_cfg = train_data_cfg['dataset']
         _retrieve_data_cfg(train_data_cfg, skip_type, show_origin)
     return cfg
 
 
 def main():
     args = parse_args()
-    cfg = retrieve_data_cfg(args.config, args.skip_type, args.show_origin)
+    cfg = retrieve_data_cfg(args.config, args.skip_type, args.cfg_options,
+                            args.show_origin)
     dataset = build_dataset(cfg.data.train)
     progress_bar = mmcv.ProgressBar(len(dataset))
     for item in dataset: