Skip to content

P
PaddleSeg

PaddlePaddle / PaddleSeg

通知 286
Star 8
Fork 1
P
PaddleSeg
提交 1dcfc05d 编写于 作者: L LutaoChu 提交者： wuzewu
浏览文件
操作

modify labelme jingling + fix label bug (#91) 

* modify annotation stript path
上级 f63d7dc0
隐藏空白更改
内联 并排
Showing with 68 addition and 69 deletion
docs/annotation/data_annotated/aa63d7e6db0d03137883772c246c6761fc201059.json 已删除 100644 → 0
浏览文件 @ f63d7dc0
{"path":"/Users/chulutao/dataset/humanseg/aa63d7e6db0d03137883772c246c6761fc201059.jpg","outputs":{"object":[{"name":"person","polygon":{"x1":321.99,"y1":63,"x2":293,"y2":98.00999999999999,"x3":245.01,"y3":141.01,"x4":221,"y4":194,"x5":231.99,"y5":237,"x6":231.99,"y6":348.01,"x7":191,"y7":429,"x8":197,"y8":465.01,"x9":193,"y9":586,"x10":151,"y10":618.01,"x11":124,"y11":622,"x12":100,"y12":703,"x13":121.99,"y13":744,"x14":141.99,"y14":724,"x15":163,"y15":658.01,"x16":238.01,"y16":646,"x17":259,"y17":627,"x18":313,"y18":618.01,"x19":416,"y19":639,"x20":464,"y20":606,"x21":454,"y21":555.01,"x22":404,"y22":508.01,"x23":430,"y23":489,"x24":407,"y24":464,"x25":397,"y25":365.01,"x26":407,"y26":290,"x27":361.99,"y27":252,"x28":376,"y28":215.01,"x29":391.99,"y29":189,"x30":388.01,"y30":135.01,"x31":340,"y31":120,"x32":313,"y32":161.01,"x33":307,"y33":188.01,"x34":311,"y34":207,"x35":277,"y35":186,"x36":293,"y36":137,"x37":308.01,"y37":117,"x38":361,"y38":93}}]},"time_labeled":1568101256852,"labeled":true,"size":{"width":706,"height":1000,"depth":3}}
\ No newline at end of file
docs/annotation/jingling2seg.md
浏览文件 @ 1dcfc05d
......@@ -44,7 +44,7 @@
**注意：导出的标注文件位于`保存位置`下的`outputs`目录。**
精灵标注产出的真值文件可参考我们给出的文件夹`data_annotated`。
精灵标注产出的真值文件可参考我们给出的文件夹`docs/annotation/jingling_demo`。
<div align="center">
<img src="../imgs/annotation/jingling-4.png" width="300px"/>
......@@ -55,21 +55,24 @@
## 3 数据格式转换
* 我们用于完成语义分割的数据集目录结构如下：
* 经过数据格式转换后的数据集目录结构如下：
```
my_dataset # 根目录
|-- JPEGImages # 数据集图片
|-- SegmentationClassPNG # 数据集真值
| |-- xxx.png # 像素级别的真值信息
| |...
|-- class_names.txt # 数据集的类别名称
|-- outputs # 标注工具导出目录
| |-- annotations # 数据集真值
| |-- xxx.png # 像素级别的真值信息
| |...
| |-- class_names.txt # 数据集的类别名称
| |-- xxx.json # 标注json文件
|-- xxx.jpg(png or other) # 数据集原图
|-- ...
```
<div align="center">
<img src="../imgs/annotation/image-6.png" width="600px"/>
<p>图5 训练所需的数据集目录的结构示意图</p>
<img src="../imgs/annotation/image-6-2.png" width="600px"/>
<p>图5 格式转换后的数据集目录的结构示意图</p>
</div>
* 运行转换脚本需要依赖labelme和pillow，如未安装，请先安装。Labelme的具体安装流程请参见[官方安装指南](https://github.com/wkentaro/labelme)。Pillow的安装：
......@@ -81,16 +84,15 @@ pip install pillow
* 运行以下代码，将标注后的数据转换成满足以上格式的数据集：
```
python docs/annotation/jingling2seg.py <path/to/label_json_file> <path/to/output_dataset>
python pdseg/tools/jingling2seg.py <path/to/label_json_file>
```
其中，`<path/to/label_json_files>`为精灵标注产出的json文件所在文件夹的目录，一般为精灵工具使用（3）中`保存位置`下的`outputs`目录。`<path/to/output_dataset>`为转换后的数据集所在文件夹的目录。
其中，`<path/to/label_json_files>`为精灵标注产出的json文件所在文件夹的目录，一般为精灵工具使用（3）中`保存位置`下的`outputs`目录。
**注意：`<path/to/output_dataset>`不用预先创建，脚本运行时会自动创建，否则会报错。**
转换得到的数据集可参考我们给出的文件夹`my_dataset`。其中，文件`class_names.txt`是数据集中所有标注类别的名称，包含背景类；文件夹`JPEGImages`保存的是数据集的图片；文件夹`SegmentationClassPNG`保存的是各图片的像素级别的真值信息，背景类`_background_`对应为0，其它目标类别从1开始递增，至多为255。
转换得到的数据集可参考我们给出的文件夹`docs/annotation/jingling_demo`。其中，文件`class_names.txt`是数据集中所有标注类别的名称，包含背景类；文件夹`annotations`保存的是各图片的像素级别的真值信息，背景类`_background_`对应为0，其它目标类别从1开始递增，至多为255。
<div align="center">
<img src="../imgs/annotation/jingling-5.png" width="600px"/>
<p>图6 训练所需的数据集各目录的内容示意图</p>
<p>图6 格式转换后的数据集各目录的内容示意图</p>
</div>
docs/annotation/jingling_demo/outputs/aa63d7e6db0d03137883772c246c6761fc201059.json 0 → 100644
浏览文件 @ 1dcfc05d
{"path":"/Users/dataset/aa63d7e6db0d03137883772c246c6761fc201059.jpg","outputs":{"object":[{"name":"person","polygon":{"x1":321.99,"y1":63,"x2":293,"y2":98.00999999999999,"x3":245.01,"y3":141.01,"x4":221,"y4":194,"x5":231.99,"y5":237,"x6":231.99,"y6":348.01,"x7":191,"y7":429,"x8":197,"y8":465.01,"x9":193,"y9":586,"x10":151,"y10":618.01,"x11":124,"y11":622,"x12":100,"y12":703,"x13":121.99,"y13":744,"x14":141.99,"y14":724,"x15":163,"y15":658.01,"x16":238.01,"y16":646,"x17":259,"y17":627,"x18":313,"y18":618.01,"x19":416,"y19":639,"x20":464,"y20":606,"x21":454,"y21":555.01,"x22":404,"y22":508.01,"x23":430,"y23":489,"x24":407,"y24":464,"x25":397,"y25":365.01,"x26":407,"y26":290,"x27":361.99,"y27":252,"x28":376,"y28":215.01,"x29":391.99,"y29":189,"x30":388.01,"y30":135.01,"x31":340,"y31":120,"x32":313,"y32":161.01,"x33":307,"y33":188.01,"x34":311,"y34":207,"x35":277,"y35":186,"x36":293,"y36":137,"x37":308.01,"y37":117,"x38":361,"y38":93}}]},"time_labeled":1568101256852,"labeled":true,"size":{"width":706,"height":1000,"depth":3}}
\ No newline at end of file
docs/annotation/jingling_demo/outputs/class_names.txt 0 → 100644
浏览文件 @ 1dcfc05d
_background_
person
\ No newline at end of file
docs/annotation/labelme2seg.md
浏览文件 @ 1dcfc05d
......@@ -47,7 +47,7 @@ git clone https://github.com/wkentaro/labelme
​ (3) 图片中所有目标的标注都完成后，点击`Save`保存json文件，**请将json文件和图片放在同一个文件夹里**，点击`Next Image`标注下一张图片。
LableMe产出的真值文件可参考我们给出的文件夹`data_annotated`。
LableMe产出的真值文件可参考我们给出的文件夹`docs/annotation/labelme_demo`。
<div align="center">
<img src="../imgs/annotation/image-5.png" width="600px"/>
......@@ -65,21 +65,24 @@ LableMe产出的真值文件可参考我们给出的文件夹`data_annotated`。
## 3 数据格式转换
* 我们用于完成语义分割的数据集目录结构如下：
* 经过数据格式转换后的数据集目录结构如下：
```
my_dataset # 根目录
|-- JPEGImages # 数据集图片
|-- SegmentationClassPNG # 数据集真值
|-- annotations # 数据集真值
| |-- xxx.png # 像素级别的真值信息
| |...
|-- class_names.txt # 数据集的类别名称
|-- xxx.jpg(png or other) # 数据集原图
|-- ...
|-- xxx.json # 标注json文件
|-- ...
```
<div align="center">
<img src="../imgs/annotation/image-6.png" width="600px"/>
<p>图7 训练所需的数据集目录的结构示意图</p>
<p>图7 格式转换后的数据集目录的结构示意图</p>
</div>
* 运行转换脚本需要依赖labelme和pillow，如未安装，请先安装。Labelme的具体安装流程请参见[官方安装指南](https://github.com/wkentaro/labelme)。Pillow的安装：
......@@ -91,14 +94,14 @@ pip install pillow
* 运行以下代码，将标注后的数据转换成满足以上格式的数据集：
```
python docs/annotation/labelme2seg.py <path/to/label_json_file> <path/to/output_dataset>
python pdseg/tools/labelme2seg.py <path/to/label_json_file>
```
其中，`<path/to/label_json_files>`为图片以及LabelMe产出的json文件所在文件夹的目录，`<path/to/output_dataset>`为转换后的数据集所在文件夹的目录。**需注意的是：`<path/to/output_dataset>`不用预先创建，脚本运行时会自动创建，否则会报错。**
其中，`<path/to/label_json_files>`为图片以及LabelMe产出的json文件所在文件夹的目录，同时也是转换后的标注集所在文件夹的目录。
转换得到的数据集可参考我们给出的文件夹`my_dataset`。其中，文件`class_names.txt`是数据集中所有标注类别的名称，包含背景类；文件夹`JPEGImages`保存的是数据集的图片；文件夹`SegmentationClassPNG`保存的是各图片的像素级别的真值信息，背景类`_background_`对应为0，其它目标类别从1开始递增，至多为255。
转换得到的数据集可参考我们给出的文件夹`docs/annotation/labelme_demo`。其中，文件`class_names.txt`是数据集中所有标注类别的名称，包含背景类；文件夹`annotations`保存的是各图片的像素级别的真值信息，背景类`_background_`对应为0，其它目标类别从1开始递增，至多为255。
<div align="center">
<img src="../imgs/annotation/image-7.png" width="600px"/>
<p>图8 训练所需的数据集各目录的内容示意图</p>
<p>图8 格式转换后的数据集各目录的内容示意图</p>
</div>
docs/imgs/annotation/image-5.png
查看替换文件 @ f63d7dc0 浏览文件 @ 1dcfc05d
docs/imgs/annotation/image-5.png

15.0 KB | W: | H:

docs/imgs/annotation/image-5.png

51.4 KB | W: | H:

docs/imgs/annotation/image-5.png
docs/imgs/annotation/image-5.png
docs/imgs/annotation/image-5.png
docs/imgs/annotation/image-5.png
  • 2-up
  • Swipe
  • Onion skin
docs/imgs/annotation/image-6.png
查看替换文件 @ f63d7dc0 浏览文件 @ 1dcfc05d
docs/imgs/annotation/image-6.png

15.2 KB | W: | H:

docs/imgs/annotation/image-6.png

77.6 KB | W: | H:

docs/imgs/annotation/image-6.png
docs/imgs/annotation/image-6.png
docs/imgs/annotation/image-6.png
docs/imgs/annotation/image-6.png
  • 2-up
  • Swipe
  • Onion skin
docs/imgs/annotation/image-7.png
查看替换文件 @ f63d7dc0 浏览文件 @ 1dcfc05d
docs/imgs/annotation/image-7.png

10.1 KB | W: | H:

docs/imgs/annotation/image-7.png

154.9 KB | W: | H:

docs/imgs/annotation/image-7.png
docs/imgs/annotation/image-7.png
docs/imgs/annotation/image-7.png
docs/imgs/annotation/image-7.png
  • 2-up
  • Swipe
  • Onion skin
docs/imgs/annotation/jingling-5.png
查看替换文件 @ f63d7dc0 浏览文件 @ 1dcfc05d
docs/imgs/annotation/jingling-5.png

153.7 KB | W: | H:

docs/imgs/annotation/jingling-5.png

112.5 KB | W: | H:

docs/imgs/annotation/jingling-5.png
docs/imgs/annotation/jingling-5.png
docs/imgs/annotation/jingling-5.png
docs/imgs/annotation/jingling-5.png
  • 2-up
  • Swipe
  • Onion skin
docs/annotation/jingling2seg.py → pdseg/tools/jingling2seg.py
浏览文件 @ 1dcfc05d
......@@ -7,7 +7,6 @@ import glob
import json
import os
import os.path as osp
import sys
import numpy as np
import PIL.Image
......@@ -15,26 +14,24 @@ import PIL.Image
import labelme
def main():
def parse_args():
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter
)
parser.add_argument('label_dir', help='input annotated directory')
parser.add_argument('output_dir', help='output dataset directory')
args = parser.parse_args()
parser.add_argument('input_dir',
help='input annotated directory')
return parser.parse_args()
if osp.exists(args.output_dir):
print('Output directory already exists:', args.output_dir)
sys.exit(1)
os.makedirs(args.output_dir)
os.makedirs(osp.join(args.output_dir, 'JPEGImages'))
os.makedirs(osp.join(args.output_dir, 'SegmentationClassPNG'))
print('Creating dataset:', args.output_dir)
def main(args):
output_dir = osp.join(args.input_dir, 'annotations')
if not osp.exists(output_dir):
os.makedirs(output_dir)
print('Creating annotations directory:', output_dir)
# get the all class names for the given dataset
class_names = ['_background_']
for label_file in glob.glob(osp.join(args.label_dir, '*.json')):
for label_file in glob.glob(osp.join(args.input_dir, '*.json')):
with open(label_file) as f:
data = json.load(f)
if data['outputs']:
......@@ -53,19 +50,17 @@ def main():
class_names = tuple(class_names)
print('class_names:', class_names)
out_class_names_file = osp.join(args.output_dir, 'class_names.txt')
out_class_names_file = osp.join(args.input_dir, 'class_names.txt')
with open(out_class_names_file, 'w') as f:
f.writelines('\n'.join(class_names))
print('Saved class_names:', out_class_names_file)
for label_file in glob.glob(osp.join(args.label_dir, '*.json')):
for label_file in glob.glob(osp.join(args.input_dir, '*.json')):
print('Generating dataset from:', label_file)
with open(label_file) as f:
base = osp.splitext(osp.basename(label_file))[0]
out_img_file = osp.join(
args.output_dir, 'JPEGImages', base + '.jpg')
out_png_file = osp.join(
args.output_dir, 'SegmentationClassPNG', base + '.png')
output_dir, base + '.png')
data = json.load(f)
......@@ -83,19 +78,18 @@ def main():
shape = {'label': name, 'points': points, 'shape_type': 'polygon'}
data_shapes.append(shape)
img_file = osp.join(osp.dirname(label_file), data['path'])
img = np.asarray(PIL.Image.open(img_file))
PIL.Image.fromarray(img).save(out_img_file)
data_size = data['size']
img_shape = (data_size['height'], data_size['width'], data_size['depth'])
lbl = labelme.utils.shapes_to_label(
img_shape=img.shape,
img_shape=img_shape,
shapes=data_shapes,
label_name_to_value=class_name_to_id,
)
if osp.splitext(out_png_file)[1] != '.png':
out_png_file += '.png'
# Assume label ranses [0, 255] for uint8,
# Assume label ranges [0, 255] for uint8,
if lbl.min() >= 0 and lbl.max() <= 255:
lbl_pil = PIL.Image.fromarray(lbl.astype(np.uint8), mode='L')
lbl_pil.save(out_png_file)
......@@ -107,4 +101,5 @@ def main():
if __name__ == '__main__':
main()
args = parse_args()
main(args)
docs/annotation/labelme2seg.py → pdseg/tools/labelme2seg.py
浏览文件 @ 1dcfc05d
......@@ -7,7 +7,6 @@ import glob
import json
import os
import os.path as osp
import sys
import numpy as np
import PIL.Image
......@@ -15,21 +14,20 @@ import PIL.Image
import labelme
def main():
def parse_args():
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter
)
parser.add_argument('input_dir', help='input annotated directory')
parser.add_argument('output_dir', help='output dataset directory')
args = parser.parse_args()
if osp.exists(args.output_dir):
print('Output directory already exists:', args.output_dir)
sys.exit(1)
os.makedirs(args.output_dir)
os.makedirs(osp.join(args.output_dir, 'JPEGImages'))
os.makedirs(osp.join(args.output_dir, 'SegmentationClassPNG'))
print('Creating dataset:', args.output_dir)
parser.add_argument('input_dir',
help='input annotated directory')
return parser.parse_args()
def main(args):
output_dir = osp.join(args.input_dir, 'annotations')
if not osp.exists(output_dir):
os.makedirs(output_dir)
print('Creating annotations directory:', output_dir)
# get the all class names for the given dataset
class_names = ['_background_']
......@@ -45,14 +43,14 @@ def main():
class_name_to_id = {}
for i, class_name in enumerate(class_names):
class_id = i # starts with 0
class_id = i # starts with 0
class_name_to_id[class_name] = class_id
if class_id == 0:
assert class_name == '_background_'
class_names = tuple(class_names)
print('class_names:', class_names)
out_class_names_file = osp.join(args.output_dir, 'class_names.txt')
out_class_names_file = osp.join(args.input_dir, 'class_names.txt')
with open(out_class_names_file, 'w') as f:
f.writelines('\n'.join(class_names))
print('Saved class_names:', out_class_names_file)
......@@ -61,16 +59,13 @@ def main():
print('Generating dataset from:', label_file)
with open(label_file) as f:
base = osp.splitext(osp.basename(label_file))[0]
out_img_file = osp.join(
args.output_dir, 'JPEGImages', base + '.jpg')
out_png_file = osp.join(
args.output_dir, 'SegmentationClassPNG', base + '.png')
output_dir, base + '.png')
data = json.load(f)
img_file = osp.join(osp.dirname(label_file), data['imagePath'])
img = np.asarray(PIL.Image.open(img_file))
PIL.Image.fromarray(img).save(out_img_file)
lbl = labelme.utils.shapes_to_label(
img_shape=img.shape,
......@@ -80,7 +75,7 @@ def main():
if osp.splitext(out_png_file)[1] != '.png':
out_png_file += '.png'
# Assume label ranses [0, 255] for uint8,
# Assume label ranges [0, 255] for uint8,
if lbl.min() >= 0 and lbl.max() <= 255:
lbl_pil = PIL.Image.fromarray(lbl.astype(np.uint8), mode='L')
lbl_pil.save(out_png_file)
......@@ -90,5 +85,7 @@ def main():
'Please consider using the .npy format.' % out_png_file
)
if __name__ == '__main__':
main()
args = parse_args()
main(args)
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或