[Fix] Replace mmcv.fileio with mmengine (#1292)

mmocr/datasets/transforms/loading.py

@@ -5,6 +5,7 @@ import warnings
 from typing import Optional
 
 import mmcv
+import mmengine
 import numpy as np
 from mmcv.transforms import BaseTransform
 from mmcv.transforms import LoadAnnotations as MMCV_LoadAnnotations
@@ -38,7 +39,7 @@ class LoadImageFromFile(MMCV_LoadImageFromFile):
             See :func:``mmcv.imfrombytes`` for details.
             Defaults to 'cv2'.
         file_client_args (dict): Arguments to instantiate a FileClient.
-            See :class:`mmcv.fileio.FileClient` for details.
+            See :class:`mmengine.fileio.FileClient` for details.
             Defaults to ``dict(backend='disk')``.
         ignore_empty (bool): Whether to allow loading empty image or file path
             not existent. Defaults to False.
@@ -59,7 +60,7 @@ class LoadImageFromFile(MMCV_LoadImageFromFile):
         self.color_type = color_type
         self.imdecode_backend = imdecode_backend
         self.file_client_args = file_client_args.copy()
-        self.file_client = mmcv.FileClient(**self.file_client_args)
+        self.file_client = mmengine.FileClient(**self.file_client_args)
         self.min_size = min_size
 
     def transform(self, results: dict) -> Optional[dict]:
@@ -473,7 +474,7 @@ class LoadImageFromLMDB(BaseTransform):
             See :func:``mmcv.imfrombytes`` for details.
             Defaults to 'cv2'.
         file_client_args (dict): Arguments to instantiate a FileClient.
-            See :class:`mmcv.fileio.FileClient` for details.
+            See :class:`mmengine.fileio.FileClient` for details.
             Defaults to ``dict(backend='lmdb', db_path='')``.
         ignore_empty (bool): Whether to allow loading empty image or file path
             not existent. Defaults to False.
@@ -490,7 +491,7 @@ class LoadImageFromLMDB(BaseTransform):
         self.color_type = color_type
         self.imdecode_backend = imdecode_backend
         self.file_client_args = file_client_args.copy()
-        self.file_client = mmcv.FileClient(**self.file_client_args)
+        self.file_client = mmengine.FileClient(**self.file_client_args)
 
     def transform(self, results: dict) -> Optional[dict]:
         """Functions to load image from LMDB file.

mmocr/engine/hooks/visualization_hook.py

@@ -3,6 +3,7 @@ import os.path as osp
 from typing import Sequence, Union
 
 import mmcv
+import mmengine
 from mmengine.hooks import Hook
 from mmengine.runner import Runner
 from mmengine.visualization import Visualizer
@@ -26,7 +27,7 @@ class VisualizationHook(Hook):
         wait_time (float): The interval of show in seconds. Defaults
             to 0.
         file_client_args (dict): Arguments to instantiate a FileClient.
-            See :class:`mmcv.fileio.FileClient` for details.
+            See :class:`mmengine.fileio.FileClient` for details.
             Defaults to ``dict(backend='disk')``.
     """
 
@@ -72,7 +73,7 @@ class VisualizationHook(Hook):
             return
 
         if self.file_client is None:
-            self.file_client = mmcv.FileClient(**self.file_client_args)
+            self.file_client = mmengine.FileClient(**self.file_client_args)
 
         # There is no guarantee that the same batch of images
         # is visualized for each evaluation.
@@ -113,7 +114,7 @@ class VisualizationHook(Hook):
             return
 
         if self.file_client is None:
-            self.file_client = mmcv.FileClient(**self.file_client_args)
+            self.file_client = mmengine.FileClient(**self.file_client_args)
 
         for output in outputs:
             img_path = output.img_path

mmocr/utils/data_converter_utils.py

@@ -2,7 +2,7 @@
 import warnings
 from typing import Dict, Sequence
 
-import mmcv
+import mmengine
 
 from mmocr.utils import is_type_list
 
@@ -122,7 +122,7 @@ def dump_ocr_data(image_infos: Sequence[Dict], out_json_name: str,
 
         out_json['data_list'].append(single_info)
 
-    mmcv.dump(out_json, out_json_name)
+    mmengine.dump(out_json, out_json_name)
 
     return out_json
 

mmocr/visualization/visualize.py

@@ -7,6 +7,7 @@ import warnings
 
 import cv2
 import mmcv
+import mmengine
 import numpy as np
 import torch
 from matplotlib import pyplot as plt
@@ -884,6 +885,6 @@ def imshow_edge(img,
             'edges': result['edges'].detach().cpu(),
             'metas': result['img_metas'][0]
         }
-        mmcv.dump(res_dic, f'{out_file}_res.pkl')
+        mmengine.dump(res_dic, f'{out_file}_res.pkl')
 
     return vis_img

tests/test_datasets/test_icdar_dataset.py

@@ -3,7 +3,7 @@ import os.path as osp
 import tempfile
 from unittest import TestCase
 
-import mmcv
+import mmengine
 
 from mmocr.datasets.icdar_dataset import IcdarDataset
 
@@ -116,7 +116,7 @@ class TestIcdarDataset(TestCase):
             categories
         }
         self.metainfo = dict(CLASSES=('text'))
-        mmcv.dump(fake_json, json_name)
+        mmengine.dump(fake_json, json_name)
 
     def test_icdar_dataset(self):
         tmp_dir = tempfile.TemporaryDirectory()

tests/test_utils/test_data_converter_utils.py

@@ -3,7 +3,7 @@ import os.path as osp
 import tempfile
 from unittest import TestCase
 
-import mmcv
+import mmengine
 
 from mmocr.utils.data_converter_utils import (dump_ocr_data,
                                               recog_anno_to_imginfo)
@@ -127,15 +127,15 @@ class TestDataConverterUtils(TestCase):
                 self._create_dummy_data()
 
             dump_ocr_data(input_data, output_path, 'textdet')
-            result = mmcv.load(output_path)
+            result = mmengine.load(output_path)
             self.assertDictEqual(result, det_target)
 
             dump_ocr_data(input_data, output_path, 'textspotter')
-            result = mmcv.load(output_path)
+            result = mmengine.load(output_path)
             self.assertDictEqual(result, spotter_target)
 
             dump_ocr_data(input_data, output_path, 'textrecog')
-            result = mmcv.load(output_path)
+            result = mmengine.load(output_path)
             self.assertDictEqual(result, recog_target)
 
     def test_recog_anno_to_imginfo(self):

tools/dataset_converters/common/curvedsyntext_converter.py

@@ -4,6 +4,7 @@ import os.path as osp
 from functools import partial
 
 import mmcv
+import mmengine
 import numpy as np
 
 from mmocr.utils import bezier_to_polygon, sort_points
@@ -104,14 +105,14 @@ def main():
     out_dir = args.out_dir if args.out_dir else root_path
     mmcv.mkdir_or_exist(out_dir)
 
-    anns = mmcv.load(osp.join(root_path, 'train1.json'))
+    anns = mmengine.load(osp.join(root_path, 'train1.json'))
     data1 = convert_annotations(anns, 'syntext_word_eng', args.num_sample,
                                 args.nproc)
 
     # Get the maximum image id from data1
     start_img_id = max(data1['images'], key=lambda x: x['id'])['id'] + 1
     start_ann_id = max(data1['annotations'], key=lambda x: x['id'])['id'] + 1
-    anns = mmcv.load(osp.join(root_path, 'train2.json'))
+    anns = mmengine.load(osp.join(root_path, 'train2.json'))
     data2 = convert_annotations(
         anns,
         'emcs_imgs',
@@ -122,7 +123,7 @@ def main():
 
     data1['images'] += data2['images']
     data1['annotations'] += data2['annotations']
-    mmcv.dump(data1, osp.join(out_dir, 'instances_training.json'))
+    mmengine.dump(data1, osp.join(out_dir, 'instances_training.json'))
 
 
 if __name__ == '__main__':

tools/dataset_converters/common/labelme_converter.py

@@ -7,6 +7,7 @@ import warnings
 from functools import partial
 
 import mmcv
+import mmengine
 
 from mmocr.utils import list_to_file
 from mmocr.utils.img_utils import crop_img, warp_img
@@ -21,7 +22,7 @@ def parse_labelme_json(json_file,
                        warp_flag=False):
     invalid_res = [[], [], []]
 
-    json_obj = mmcv.load(json_file)
+    json_obj = mmengine.load(json_file)
 
     img_file = osp.basename(json_obj['imagePath'])
     img_full_path = osp.join(img_dir, img_file)

tools/dataset_converters/textdet/art_converter.py

@@ -4,6 +4,7 @@ import math
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils import convert_annotations
 
@@ -61,7 +62,7 @@ def collect_art_info(root_path, split, ratio, print_every=1000):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     img_prefixes = annotation.keys()
 
     trn_files, val_files = [], []

tools/dataset_converters/textdet/coco_to_line_dict.py

@@ -2,13 +2,13 @@
 import argparse
 import json
 
-import mmcv
+import mmengine
 
 from mmocr.utils import list_to_file
 
 
 def parse_coco_json(in_path):
-    json_obj = mmcv.load(in_path)
+    json_obj = mmengine.load(in_path)
     image_infos = json_obj['images']
     annotations = json_obj['annotations']
     imgid2imgname = {}

tools/dataset_converters/textdet/cocotext_converter.py

@@ -3,7 +3,7 @@ import argparse
 import math
 import os.path as osp
 
-import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -66,7 +66,7 @@ def collect_cocotext_info(root_path, split, print_every=1000):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
 
     img_infos = []
     for i, img_info in enumerate(annotation['imgs'].values()):

tools/dataset_converters/textdet/data_migrator.py

@@ -4,7 +4,7 @@ from collections import defaultdict
 from copy import deepcopy
 from typing import Dict, List
 
-import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -19,7 +19,7 @@ def parse_coco_json(in_path: str) -> List[Dict]:
         list[dict]: List of image information dicts. To be used by
         dump_ocr_data().
     """
-    json_obj = mmcv.load(in_path)
+    json_obj = mmengine.load(in_path)
     image_infos = json_obj['images']
     annotations = json_obj['annotations']
     imgid2annos = defaultdict(list)

tools/dataset_converters/textdet/funsd_converter.py

@@ -5,6 +5,7 @@ import os
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -100,7 +101,7 @@ def load_json_info(gt_file, img_info):
         img_info (dict): The dict of the img and annotation information
     """
 
-    annotation = mmcv.load(gt_file)
+    annotation = mmengine.load(gt_file)
     anno_info = []
     for form in annotation['form']:
         for ann in form['words']:

tools/dataset_converters/textdet/imgur_converter.py

@@ -4,6 +4,7 @@ import math
 import os.path as osp
 
 import mmcv
+import mmengine
 import numpy as np
 
 from mmocr.utils import dump_ocr_data
@@ -25,7 +26,7 @@ def collect_imgur_info(root_path, annotation_filename, print_every=1000):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     images = annotation['index_to_ann_map'].keys()
     img_infos = []
     for i, img_name in enumerate(images):

tools/dataset_converters/textdet/lsvt_converter.py

@@ -4,6 +4,7 @@ import math
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -50,7 +51,7 @@ def collect_lsvt_info(root_path, split, ratio, print_every=1000):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     img_prefixes = annotation.keys()
 
     trn_files, val_files = [], []

tools/dataset_converters/textdet/naf_converter.py

@@ -3,6 +3,7 @@ import argparse
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -127,7 +128,7 @@ def load_json_info(gt_file, img_info):
     assert isinstance(gt_file, str)
     assert isinstance(img_info, dict)
 
-    annotation = mmcv.load(gt_file)
+    annotation = mmengine.load(gt_file)
     anno_info = []
 
     # 'textBBs' contains the printed texts of the table while 'fieldBBs'
@@ -175,7 +176,7 @@ def parse_args():
 def main():
     args = parse_args()
     root_path = args.root_path
-    split_info = mmcv.load(
+    split_info = mmengine.load(
         osp.join(root_path, 'annotations', 'train_valid_test_split.json'))
     split_info['training'] = split_info.pop('train')
     split_info['val'] = split_info.pop('valid')

tools/dataset_converters/textdet/rects_converter.py

@@ -5,6 +5,7 @@ import os
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -142,7 +143,7 @@ def load_json_info(gt_file, img_info):
         img_info (dict): The dict of the img and annotation information
     """
 
-    annotation = mmcv.load(gt_file)
+    annotation = mmengine.load(gt_file)
     anno_info = []
     for line in annotation['lines']:
         segmentation = line['points']

tools/dataset_converters/textdet/textocr_converter.py

@@ -3,7 +3,7 @@ import argparse
 import math
 import os.path as osp
 
-import mmcv
+import mmengine
 
 from mmocr.utils import dump_ocr_data
 
@@ -23,7 +23,7 @@ def collect_textocr_info(root_path, annotation_filename, print_every=1000):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
 
     # img_idx = img_start_idx
     img_infos = []

tools/dataset_converters/textrecog/art_converter.py

@@ -4,7 +4,7 @@ import json
 import math
 import os.path as osp
 
-import mmcv
+import mmengine
 
 from mmocr.utils.fileio import list_to_file
 
@@ -63,7 +63,7 @@ def convert_art(root_path, split, ratio, format):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     # outputs
     dst_label_file = osp.join(root_path, f'{split}_label.{format}')
 

tools/dataset_converters/textrecog/cocotext_converter.py

@@ -6,6 +6,7 @@ import os.path as osp
 from functools import partial
 
 import mmcv
+import mmengine
 
 from mmocr.utils.fileio import list_to_file
 
@@ -131,7 +132,7 @@ def convert_cocotext(root_path,
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     # outputs
     dst_label_file = osp.join(root_path, f'{split}_label.{format}')
     dst_image_root = osp.join(root_path, 'crops', split)

tools/dataset_converters/textrecog/funsd_converter.py

@@ -6,6 +6,7 @@ import os
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils.fileio import list_to_file
 from mmocr.utils.img_utils import crop_img
@@ -102,7 +103,7 @@ def load_json_info(gt_file, img_info):
         img_info (dict): The dict of the img and annotation information
     """
 
-    annotation = mmcv.load(gt_file)
+    annotation = mmengine.load(gt_file)
     anno_info = []
     for form in annotation['form']:
         for ann in form['words']:

tools/dataset_converters/textrecog/imgur_converter.py

@@ -6,6 +6,7 @@ import os
 import os.path as osp
 
 import mmcv
+import mmengine
 import numpy as np
 
 from mmocr.utils.fileio import list_to_file
@@ -33,7 +34,7 @@ def collect_imgur_info(root_path, annotation_filename, print_every=1000):
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     images = annotation['index_to_ann_map'].keys()
     img_infos = []
     for i, img_name in enumerate(images):

tools/dataset_converters/textrecog/lsvt_converter.py

@@ -6,6 +6,7 @@ import os.path as osp
 from functools import partial
 
 import mmcv
+import mmengine
 
 from mmocr.utils.fileio import list_to_file
 
@@ -114,7 +115,7 @@ def convert_lsvt(root_path,
         raise Exception(
             f'{annotation_path} not exists, please check and try again.')
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
     # outputs
     dst_label_file = osp.join(root_path, f'{split}_label.{format}')
     dst_image_root = osp.join(root_path, 'crops', split)

tools/dataset_converters/textrecog/naf_converter.py

@@ -4,6 +4,7 @@ import json
 import os.path as osp
 
 import mmcv
+import mmengine
 import numpy as np
 
 from mmocr.utils.fileio import list_to_file
@@ -138,7 +139,7 @@ def load_json_info(gt_file, img_info):
     assert isinstance(gt_file, str)
     assert isinstance(img_info, dict)
 
-    annotation = mmcv.load(gt_file)
+    annotation = mmengine.load(gt_file)
     anno_info = []
 
     # 'textBBs' contains the printed texts of the table while 'fieldBBs'
@@ -267,7 +268,7 @@ def parse_args():
 def main():
     args = parse_args()
     root_path = args.root_path
-    split_info = mmcv.load(
+    split_info = mmengine.load(
         osp.join(root_path, 'annotations', 'train_valid_test_split.json'))
     split_info['training'] = split_info.pop('train')
     split_info['val'] = split_info.pop('valid')

tools/dataset_converters/textrecog/openvino_converter.py

@@ -6,6 +6,7 @@ from argparse import ArgumentParser
 from functools import partial
 
 import mmcv
+import mmengine
 from PIL import Image
 
 from mmocr.utils.fileio import list_to_file
@@ -67,7 +68,7 @@ def convert_openimages(root_path,
     dst_image_root = osp.join(root_path, dst_image_path)
     os.makedirs(dst_image_root, exist_ok=True)
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
 
     process_img_with_path = partial(
         process_img,

tools/dataset_converters/textrecog/rects_converter.py

@@ -6,6 +6,7 @@ import os
 import os.path as osp
 
 import mmcv
+import mmengine
 
 from mmocr.utils.fileio import list_to_file
 from mmocr.utils.img_utils import crop_img
@@ -144,7 +145,7 @@ def load_json_info(gt_file, img_info):
         img_info (dict): The dict of the img and annotation information
     """
 
-    annotation = mmcv.load(gt_file)
+    annotation = mmengine.load(gt_file)
     anno_info = []
     for line in annotation['lines']:
         if line['ignore'] == 1:

tools/dataset_converters/textrecog/textocr_converter.py

@@ -6,6 +6,7 @@ import os.path as osp
 from functools import partial
 
 import mmcv
+import mmengine
 
 from mmocr.utils.fileio import list_to_file
 
@@ -63,7 +64,7 @@ def convert_textocr(root_path,
     dst_image_root = osp.join(root_path, dst_image_path)
     os.makedirs(dst_image_root, exist_ok=True)
 
-    annotation = mmcv.load(annotation_path)
+    annotation = mmengine.load(annotation_path)
 
     process_img_with_path = partial(
         process_img,