Merge pull request open-mmlab#16 from opencv/reading_annotation_files…

…_by_pattern

Reading annotation files by pattern

mmdet/datasets/builder.py

@@ -1,6 +1,9 @@
 import copy
+import glob
+import os
 
 from mmdet.utils import build_from_cfg
+
 from .dataset_wrappers import ConcatDataset, RepeatDataset
 from .registry import DATASETS
 
@@ -11,6 +14,25 @@ def _concat_dataset(cfg, default_args=None):
     seg_prefixes = cfg.get('seg_prefix', None)
     proposal_files = cfg.get('proposal_file', None)
 
+    if cfg.get('img_prefix_auto', False):
+        del cfg['img_prefix_auto']
+        assert img_prefixes is None
+        if cfg['type'] == 'CustomCocoDataset':
+            # assuming following dataset structure:
+            # dataset_root
+            # ├── annotations
+            # │   ├── instances_train.json
+            # │   ├── ...
+            # ├── images
+            #     ├── image_name1
+            #     ├── image_name2
+            #     ├── ...
+            # and file_name inside instances_train.json is relative to <dataset_root>/images
+            img_prefixes = \
+                [os.path.join(os.path.dirname(ann_file), '..', 'images') for ann_file in ann_files]
+        else:
+            raise NotImplementedError
+
     datasets = []
     num_dset = len(ann_files)
     for i in range(num_dset):
@@ -36,6 +58,15 @@ def build_dataset(cfg, default_args=None):
     elif isinstance(cfg['ann_file'], (list, tuple)):
         dataset = _concat_dataset(cfg, default_args)
     else:
-        dataset = build_from_cfg(cfg, DATASETS, default_args)
+        matches = glob.glob(cfg['ann_file'], recursive=True)
+        if not matches:
+            raise RuntimeError(f'Failed to find annotation files that match pattern: '
+                               f'{cfg["ann_file"]}')
+        cfg['ann_file'] = matches
+        if len(cfg['ann_file']) == 1:
+            cfg['ann_file'] = cfg['ann_file'][0]
+            dataset = build_from_cfg(cfg, DATASETS, default_args)
+        else:
+            dataset = _concat_dataset(cfg, default_args)
 
     return dataset