mirrors
/
mmocr
mirror of https://github.com/open-mmlab/mmocr.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[Fix] Fix TotalText Anno version issue (#945) 

* fix tt converter version issue; fix typos in docs

* remove incorrect descriptions

* fix docstring & incorrect file name

* fix docstring identation
pull/971/head
Xinyu Wang 2022-04-23 23:57:21 +08:00 committed by GitHub
parent e51d8533ea
commit 06b73cf71a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 153 additions and 153 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
docs/en/datasets/det.md
View File

@ -255,7 +255,7 @@ inconsistency results in false examples in the training set. Therefore, users sh
## Totaltext
- Step0: Read [Important Note](#important-note)
- Step1: Download `totaltext.zip` from [github dataset](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Dataset) and `groundtruth_text.zip` from [github Groundtruth](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Groundtruth/Text) (Our totaltext_converter.py supports groundtruth with both .mat and .txt format).
- Step1: Download `totaltext.zip` from [github dataset](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Dataset) and `groundtruth_text.zip` or `TT_new_train_GT.zip` (if you prefer to use the latest version of training annotations) from [github Groundtruth](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Groundtruth/Text) (Our totaltext_converter.py supports groundtruth with both .mat and .txt format).
```bash
mkdir totaltext && cd totaltext
@ -267,17 +267,21 @@ inconsistency results in false examples in the training set. Therefore, users sh
mv Images/Train imgs/training
mv Images/Test imgs/test
# For annotations
# For legacy training and test annotations
unzip groundtruth_text.zip
cd Groundtruth
mv Polygon/Train ../annotations/training
mv Polygon/Test ../annotations/test
mv Groundtruth/Polygon/Train annotations/training
mv Groundtruth/Polygon/Test annotations/test
# Using the latest training annotations
# WARNING: Delete legacy train annotations before running the following command.
unzip TT_new_train_GT.zip
mv Train annotations/training
```
- Step2: Generate `instances_training.json` and `instances_test.json` with the following command:
```bash
python tools/data/textdet/totaltext_converter.py /path/to/totaltext -o /path/to/totaltext --split-list training test
python tools/data/textdet/totaltext_converter.py /path/to/totaltext
```
- The resulting directory structure looks like the following:
@ -507,7 +511,7 @@ inconsistency results in false examples in the training set. Therefore, users sh
│   └── instances_val.json
```
### LSVT
## LSVT
- Step1: Download [train_full_images_0.tar.gz](https://dataset-bj.cdn.bcebos.com/lsvt/train_full_images_0.tar.gz), [train_full_images_1.tar.gz](https://dataset-bj.cdn.bcebos.com/lsvt/train_full_images_1.tar.gz), and [train_full_labels.json](https://dataset-bj.cdn.bcebos.com/lsvt/train_full_labels.json) to `lsvt/`.
@ -705,8 +709,6 @@ inconsistency results in false examples in the training set. Therefore, users sh
```bash
# Annotations of ReCTS test split is not publicly available, split a validation
# set by adding --val-ratio 0.2
# Add --preserve-vertical to preserve vertical texts for training, otherwise
# vertical images will be filtered and stored in PATH/TO/rects/ignores
python tools/data/textdet/rects_converter.py PATH/TO/rects --nproc 4 --val-ratio 0.2
```
@ -853,11 +855,10 @@ inconsistency results in false examples in the training set. Therefore, users sh
- Step1: Download `train_images.zip.001`, `train_images.zip.002`, and `train_gts.zip` from the [homepage](https://rctw.vlrlab.net/dataset.html), extract the zips to `rctw/imgs` and `rctw/annotations`, respectively.
- Step2: Generate `instances_training.json` and `instances_val.json` (optional). Since the original dataset doesn't have a validation set, you may specify `--val-ratio` to split the dataset. E.g., if val-ratio is 0.2, then 20% of the data are left out as the validation set in this example.
- Step2: Generate `instances_training.json` and `instances_val.json` (optional). Since the test annotations are not publicly available, you may specify `--val-ratio` to split the dataset. E.g., if val-ratio is 0.2, then 20% of the data are left out as the validation set in this example.
```bash
# Annotations of RCTW test split is not publicly available, split a validation set by adding --val-ratio 0.2
# Add --preserve-vertical to preserve vertical texts for training, otherwise vertical images will be filtered and stored in PATH/TO/rctw/ignores
python tools/data/textdet/rctw_converter.py PATH/TO/rctw --nproc 4
```

29
docs/en/datasets/recog.md
View File

@ -392,7 +392,7 @@ should be as follows:
## Totaltext
- Step1: Download `totaltext.zip` from [github dataset](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Dataset) and `groundtruth_text.zip` from [github Groundtruth](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Groundtruth/Text) (Our totaltext_converter.py supports groundtruth with both .mat and .txt format).
- Step1: Download `totaltext.zip` from [github dataset](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Dataset) and `groundtruth_text.zip` or `TT_new_train_GT.zip` (if you prefer to use the latest version of training annotations) from [github Groundtruth](https://github.com/cs-chan/Total-Text-Dataset/tree/master/Groundtruth/Text) (Our totaltext_converter.py supports groundtruth with both .mat and .txt format).
```bash
mkdir totaltext && cd totaltext
@ -404,27 +404,28 @@ should be as follows:
mv Images/Train imgs/training
mv Images/Test imgs/test
# For annotations
# For legacy training and test annotations
unzip groundtruth_text.zip
cd Groundtruth
mv Polygon/Train ../annotations/training
mv Polygon/Test ../annotations/test
mv Groundtruth/Polygon/Train annotations/training
mv Groundtruth/Polygon/Test annotations/test
# Using the latest training annotations
# WARNING: Delete legacy train annotations before running the following command.
unzip TT_new_train_GT.zip
mv Train annotations/training
```
- Step2: Generate cropped images, `train_label.txt` and `test_label.txt` with the following command (the cropped images will be saved to `data/totaltext/dst_imgs/`):
```bash
python tools/data/textrecog/totaltext_converter.py /path/to/totaltext -o /path/to/totaltext --split-list training test
python tools/data/textrecog/totaltext_converter.py /path/to/totaltext
```
- After running the above codes, the directory structure
should be as follows:
- After running the above codes, the directory structure should be as follows:
```text
├── Totaltext
│ ├── imgs
│ ├── annotations
├── totaltext
│ ├── dst_imgs
│ ├── train_label.txt
│ └── test_label.txt
```
@ -635,7 +636,7 @@ The LV dataset has already provided cropped images and the corresponding annotat
│ └── test_label.jsonl
```
### LSVT
## LSVT
- Step1: Download [train_full_images_0.tar.gz](https://dataset-bj.cdn.bcebos.com/lsvt/train_full_images_0.tar.gz), [train_full_images_1.tar.gz](https://dataset-bj.cdn.bcebos.com/lsvt/train_full_images_1.tar.gz), and [train_full_labels.json](https://dataset-bj.cdn.bcebos.com/lsvt/train_full_labels.json) to `lsvt/`.
@ -655,7 +656,7 @@ The LV dataset has already provided cropped images and the corresponding annotat
rm train_full_images_0.tar.gz && rm train_full_images_1.tar.gz && rm -rf train_full_images_1
```
- Step2: Generate `train_label.jsonl` and `val_label.jsonl` (optional) with the following command:
- Step2: Generate `train_label.jsonl` and `val_label.jsonl` (optional) with the following command:
```bash
# Annotations of LSVT test split is not publicly available, split a validation
@ -672,7 +673,7 @@ The LV dataset has already provided cropped images and the corresponding annotat
│ ├── crops
│ ├── ignores
│ ├── train_label.jsonl
│ ├── val_label.jsonl (optional)
│ └── val_label.jsonl (optional)
```
## FUNSD

121
tools/data/textdet/totaltext_converter.py
View File

@ -15,15 +15,15 @@ from shapely.geometry import Polygon
from mmocr.utils import convert_annotations
def collect_files(img_dir, gt_dir, split):
def collect_files(img_dir, gt_dir):
"""Collect all images and their corresponding groundtruth files.
Args:
img_dir(str): The image directory
gt_dir(str): The groundtruth directory
split(str): The split of dataset. Namely: training or test
img_dir (str): The image directory
gt_dir (str): The groundtruth directory
Returns:
files(list): The list of tuples (img_file, groundtruth_file)
files (list): The list of tuples (img_file, groundtruth_file)
"""
assert isinstance(img_dir, str)
assert img_dir
@ -54,10 +54,11 @@ def collect_annotations(files, nproc=1):
"""Collect the annotation information.
Args:
files(list): The list of tuples (image_file, groundtruth_file)
nproc(int): The number of process to collect annotations
files (list): The list of tuples (image_file, groundtruth_file)
nproc (int): The number of process to collect annotations
Returns:
images(list): The list of image information dicts
images (list): The list of image information dicts
"""
assert isinstance(files, list)
assert isinstance(nproc, int)
@ -75,12 +76,13 @@ def get_contours_mat(gt_path):
"""Get the contours and words for each ground_truth mat file.
Args:
gt_path(str): The relative path of the ground_truth mat file
gt_path (str): The relative path of the ground_truth mat file
Returns:
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
"""
assert isinstance(gt_path, str)
@ -88,7 +90,13 @@ def get_contours_mat(gt_path):
words = []
data = scio.loadmat(gt_path)
# 'gt' for the latest version; 'polygt' for the legacy version
data_polygt = data.get('polygt', data['gt'])
keys = data.keys()
if 'gt' in keys:
data_polygt = data.get('gt')
elif 'polygt' in keys:
data_polygt = data.get('polygt')
else:
raise NotImplementedError
for i, lines in enumerate(data_polygt):
X = np.array(lines[1])
@ -96,15 +104,11 @@ def get_contours_mat(gt_path):
point_num = len(X[0])
word = lines[4]
if len(word) == 0:
word = '???'
if len(word) == 0 or word == '#':
word = '###'
else:
word = word[0]
if word == '#':
word = '###'
continue
words.append(word)
arr = np.concatenate([X, Y]).T
@ -121,9 +125,10 @@ def load_mat_info(img_info, gt_file):
"""Load the information of one ground truth in .mat format.
Args:
img_info(dict): The dict of only the image information
gt_file(str): The relative path of the ground_truth mat
file for one image
img_info (dict): The dict of only the image information
gt_file (str): The relative path of the ground_truth mat
file for one image
Returns:
img_info(dict): The dict of the img and annotation information
"""
@ -138,7 +143,7 @@ def load_mat_info(img_info, gt_file):
category_id = 1
coordinates = np.array(contour).reshape(-1, 2)
polygon = Polygon(coordinates)
iscrowd = 0
iscrowd = 1 if text == '###' else 0
area = polygon.area
# convert to COCO style XYWH format
@ -165,14 +170,15 @@ def process_line(line, contours, words):
Args:
line(str): The line in gt file containing annotation info
contours(list[lists]): A list of lists of contours
for the text instances
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
for the text instances
Returns:
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
"""
line = '{' + line.replace('[[', '[').replace(']]', ']') + '}'
@ -186,7 +192,7 @@ def process_line(line, contours, words):
Y = np.array([ann_dict['y']])
if len(ann_dict['transcriptions']) == 0:
word = '???'
word = '###'
else:
word = ann_dict['transcriptions'][0]
if len(ann_dict['transcriptions']) > 1:
@ -211,12 +217,13 @@ def get_contours_txt(gt_path):
"""Get the contours and words for each ground_truth txt file.
Args:
gt_path(str): The relative path of the ground_truth mat file
gt_path (str): The relative path of the ground_truth mat file
Returns:
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
"""
assert isinstance(gt_path, str)
@ -250,9 +257,10 @@ def load_txt_info(gt_file, img_info):
"""Load the information of one ground truth in .txt format.
Args:
img_info(dict): The dict of only the image information
gt_file(str): The relative path of the ground_truth mat
file for one image
img_info (dict): The dict of only the image information
gt_file (str): The relative path of the ground_truth mat
file for one image
Returns:
img_info(dict): The dict of the img and annotation information
"""
@ -265,7 +273,7 @@ def load_txt_info(gt_file, img_info):
category_id = 1
coordinates = np.array(contour).reshape(-1, 2)
polygon = Polygon(coordinates)
iscrowd = 0
iscrowd = 1 if text == '###' else 0
area = polygon.area
# convert to COCO style XYWH format
@ -290,10 +298,11 @@ def load_png_info(gt_file, img_info):
"""Load the information of one ground truth in .png format.
Args:
gt_file(str): The relative path of the ground_truth file for one image
img_info(dict): The dict of only the image information
gt_file (str): The relative path of the ground_truth file for one image
img_info (dict): The dict of only the image information
Returns:
img_info(dict): The dict of the img and annotation information
img_info (dict): The dict of the img and annotation information
"""
assert isinstance(gt_file, str)
assert isinstance(img_info, dict)
@ -334,14 +343,15 @@ def load_img_info(files):
"""Load the information of one image.
Args:
files(tuple): The tuple of (img_file, groundtruth_file)
files (tuple): The tuple of (img_file, groundtruth_file)
Returns:
img_info(dict): The dict of the img and annotation information
img_info (dict): The dict of the img and annotation information
"""
assert isinstance(files, tuple)
img_file, gt_file = files
# read imgs with ignoring orientations
# read imgs while ignoring orientations
img = mmcv.imread(img_file, 'unchanged')
split_name = osp.basename(osp.dirname(img_file))
@ -366,15 +376,9 @@ def load_img_info(files):
def parse_args():
parser = argparse.ArgumentParser(
description='Convert totaltext annotations to COCO format')
parser.add_argument('root_path', help='totaltext root path')
parser.add_argument('-o', '--out-dir', help='output path')
parser.add_argument('root_path', help='Totaltext root path')
parser.add_argument(
'--split-list',
nargs='+',
help='a list of splits. e.g., "--split_list training test"')
parser.add_argument(
'--nproc', default=1, type=int, help='number of process')
'--nproc', default=1, type=int, help='Number of process')
args = parser.parse_args()
return args
@ -382,14 +386,11 @@ def parse_args():
def main():
args = parse_args()
root_path = args.root_path
out_dir = args.out_dir if args.out_dir else root_path
mmcv.mkdir_or_exist(out_dir)
img_dir = osp.join(root_path, 'imgs')
gt_dir = osp.join(root_path, 'annotations')
set_name = {}
for split in args.split_list:
for split in ['training', 'test']:
set_name.update({split: 'instances_' + split + '.json'})
assert osp.exists(osp.join(img_dir, split))
@ -398,9 +399,9 @@ def main():
with mmcv.Timer(
print_tmpl='It takes {}s to convert totaltext annotation'):
files = collect_files(
osp.join(img_dir, split), osp.join(gt_dir, split), split)
osp.join(img_dir, split), osp.join(gt_dir, split))
image_infos = collect_annotations(files, nproc=args.nproc)
convert_annotations(image_infos, osp.join(out_dir, json_name))
convert_annotations(image_infos, osp.join(root_path, json_name))
if __name__ == '__main__':

133
tools/data/textrecog/totaltext_converter.py
View File

@ -15,13 +15,13 @@ from mmocr.datasets.pipelines.crop import crop_img
from mmocr.utils.fileio import list_to_file
def collect_files(img_dir, gt_dir, split):
def collect_files(img_dir, gt_dir):
"""Collect all images and their corresponding groundtruth files.
Args:
img_dir(str): The image directory
gt_dir(str): The groundtruth directory
split(str): The split of dataset. Namely: training or test
img_dir (str): The image directory
gt_dir (str): The groundtruth directory
Returns:
files(list): The list of tuples (img_file, groundtruth_file)
"""
@ -55,10 +55,11 @@ def collect_annotations(files, nproc=1):
"""Collect the annotation information.
Args:
files(list): The list of tuples (image_file, groundtruth_file)
nproc(int): The number of process to collect annotations
files (list): The list of tuples (image_file, groundtruth_file)
nproc (int): The number of process to collect annotations
Returns:
images(list): The list of image information dicts
images (list): The list of image information dicts
"""
assert isinstance(files, list)
assert isinstance(nproc, int)
@ -76,19 +77,25 @@ def get_contours_mat(gt_path):
"""Get the contours and words for each ground_truth mat file.
Args:
gt_path(str): The relative path of the ground_truth mat file
gt_path (str): The relative path of the ground_truth mat file
Returns:
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
"""
assert isinstance(gt_path, str)
contours = []
words = []
data = scio.loadmat(gt_path)
data_polygt = data['polygt']
# 'gt' for the latest version; 'polygt' for the legacy version
keys = data.keys()
if 'gt' in keys:
data_polygt = data.get('gt')
elif 'polygt' in keys:
data_polygt = data.get('polygt')
for i, lines in enumerate(data_polygt):
X = np.array(lines[1])
@ -96,15 +103,11 @@ def get_contours_mat(gt_path):
point_num = len(X[0])
word = lines[4]
if len(word) == 0:
word = '???'
if len(word) == 0 or word == '#':
word = '###'
else:
word = word[0]
if word == '#':
word = '###'
continue
words.append(word)
arr = np.concatenate([X, Y]).T
@ -121,9 +124,10 @@ def load_mat_info(img_info, gt_file):
"""Load the information of one ground truth in .mat format.
Args:
img_info(dict): The dict of only the image information
gt_file(str): The relative path of the ground_truth mat
file for one image
img_info (dict): The dict of only the image information
gt_file (str): The relative path of the ground_truth mat
file for one image
Returns:
img_info(dict): The dict of the img and annotation information
"""
@ -133,7 +137,7 @@ def load_mat_info(img_info, gt_file):
contours, words = get_contours_mat(gt_file)
anno_info = []
for contour, word in zip(contours, words):
if contour.shape[0] == 2:
if contour.shape[0] == 2 or word == '###':
continue
coordinates = np.array(contour).reshape(-1, 2)
polygon = Polygon(coordinates)
@ -152,16 +156,17 @@ def process_line(line, contours, words):
"""Get the contours and words by processing each line in the gt file.
Args:
line(str): The line in gt file containing annotation info
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
line (str): The line in gt file containing annotation info
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
Returns:
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
"""
line = '{' + line.replace('[[', '[').replace(']]', ']') + '}'
@ -175,7 +180,7 @@ def process_line(line, contours, words):
Y = np.array([ann_dict['y']])
if len(ann_dict['transcriptions']) == 0:
word = '???'
word = '###'
else:
word = ann_dict['transcriptions'][0]
if len(ann_dict['transcriptions']) > 1:
@ -200,12 +205,13 @@ def get_contours_txt(gt_path):
"""Get the contours and words for each ground_truth txt file.
Args:
gt_path(str): The relative path of the ground_truth mat file
gt_path (str): The relative path of the ground_truth mat file
Returns:
contours(list[lists]): A list of lists of contours
for the text instances
words(list[list]): A list of lists of words (string)
for the text instances
contours (list[lists]): A list of lists of contours
for the text instances
words (list[list]): A list of lists of words (string)
for the text instances
"""
assert isinstance(gt_path, str)
@ -231,10 +237,8 @@ def get_contours_txt(gt_path):
contours, words = process_line(tmp_line, contours, words)
for word in words:
if word == '#':
word = '###'
continue
return contours, words
@ -243,17 +247,18 @@ def load_txt_info(gt_file, img_info):
"""Load the information of one ground truth in .txt format.
Args:
img_info(dict): The dict of only the image information
gt_file(str): The relative path of the ground_truth mat
file for one image
img_info (dict): The dict of only the image information
gt_file (str): The relative path of the ground_truth mat
file for one image
Returns:
img_info(dict): The dict of the img and annotation information
img_info (dict): The dict of the img and annotation information
"""
contours, words = get_contours_txt(gt_file)
anno_info = []
for contour, word in zip(contours, words):
if contour.shape[0] == 2:
if contour.shape[0] == 2 or word == '###':
continue
coordinates = np.array(contour).reshape(-1, 2)
polygon = Polygon(coordinates)
@ -272,10 +277,10 @@ def generate_ann(root_path, split, image_infos):
"""Generate cropped annotations and label txt file.
Args:
root_path(str): The relative path of the totaltext file
split(str): The split of dataset. Namely: training or test
image_infos(list[dict]): A list of dicts of the img and
annotation information
root_path (str): The relative path of the totaltext file
split (str): The split of dataset. Namely: training or test
image_infos (list[dict]): A list of dicts of the img and
annotation information
"""
dst_image_root = osp.join(root_path, 'dst_imgs', split)
@ -297,7 +302,7 @@ def generate_ann(root_path, split, image_infos):
dst_img = crop_img(image, anno['bbox'])
# Skip invalid annotations
if min(dst_img.shape) == 0:
if min(dst_img.shape) == 0 or word == '###':
continue
dst_img_name = f'{src_img_root}_{index}.png'
@ -313,9 +318,10 @@ def load_img_info(files):
"""Load the information of one image.
Args:
files(tuple): The tuple of (img_file, groundtruth_file)
files (tuple): The tuple of (img_file, groundtruth_file)
Returns:
img_info(dict): The dict of the img and annotation information
img_info (dict): The dict of the img and annotation information
"""
assert isinstance(files, tuple)
@ -345,15 +351,9 @@ def load_img_info(files):
def parse_args():
parser = argparse.ArgumentParser(
description='Convert totaltext annotations to COCO format')
parser.add_argument('root_path', help='totaltext root path')
parser.add_argument('-o', '--out-dir', help='output path')
parser.add_argument('root_path', help='Totaltext root path')
parser.add_argument(
'--split-list',
nargs='+',
help='a list of splits. e.g., "--split_list training test"')
parser.add_argument(
'--nproc', default=1, type=int, help='number of process')
'--nproc', default=1, type=int, help='Number of process')
args = parser.parse_args()
return args
@ -361,23 +361,20 @@ def parse_args():
def main():
args = parse_args()
root_path = args.root_path
out_dir = args.out_dir if args.out_dir else root_path
mmcv.mkdir_or_exist(out_dir)
img_dir = osp.join(root_path, 'imgs')
gt_dir = osp.join(root_path, 'annotations')
set_name = {}
for split in args.split_list:
set_name.update({split: 'instances_' + split + '.json'})
for split in ['training', 'test']:
set_name.update({split: split + '_label' + '.txt'})
assert osp.exists(osp.join(img_dir, split))
for split, json_name in set_name.items():
print(f'Converting {split} into {json_name}')
for split, ann_name in set_name.items():
print(f'Converting {split} into {ann_name}')
with mmcv.Timer(
print_tmpl='It takes {}s to convert totaltext annotation'):
files = collect_files(
osp.join(img_dir, split), osp.join(gt_dir, split), split)
osp.join(img_dir, split), osp.join(gt_dir, split))
image_infos = collect_annotations(files, nproc=args.nproc)
generate_ann(root_path, split, image_infos)