将字符串字典类型转换为字典

Question

我正在尝试使用 python 将数据帧的列转换为 dict。

dataframe 的形式为

filename        bbox-coordinates

image_name1     '{""name"":""rect"",""x"":161,""y"":562,""width"":51,""height"":96}'
image_name2     '{""name"":""rect"",""x"":151,""y"":542,""width"":32,""height"":69}'
image_name3     '{""name"":""rect"",""x"":140,""y"":42,""width"":30,""height"":45}'

bbox 坐标值是字符串类型，我正在尝试将其转换为 dict。

dataframe['bbox-coordinate'][2]
output: '{""name"":""rect"",""x"":140,""y"":42,""width"":30,""height"":45}' which is string

这里：

 [1] type(dataframe['region_shape_attributes'][0])
output: str

我尝试使用：

ast.literal_eval --- doesn't work 
json.loads(dataframe['bbox-coordinate'][2]) 
output: JSONDecodeError: Expecting ':' delimiter: line 1 column 4 (char 3)

我的最终目标是将整个 dataframe 转换为带有图像 class 名称的 YOLO format.txt，这只有一个 class 即 0，并以 Y_O 格式提取每个文件的高度 x、宽度和宽度。

Answer 1

出于某种原因，您的字符串中似乎有双双引号（ "" ）。 只需用单个双引号（ " ）替换它们， ast.literal_eval()就可以了。

from ast import literal_eval

strings = [
    '{""name"":""rect"",""x"":161,""y"":562,""width"":51,""height"":96}',
    '{""name"":""rect"",""x"":151,""y"":542,""width"":32,""height"":69}',
    '{""name"":""rect"",""x"":140,""y"":42,""width"":30,""height"":45}'
    ]

dicts = []
for s in strings:
    d = literal_eval(s.replace('""', '"'))
    dicts.append(d)

print(dicts)

返回

[{'name': 'rect', 'x': 161, 'y': 562, 'width': 51, 'height': 96}, 
 {'name': 'rect', 'x': 151, 'y': 542, 'width': 32, 'height': 69}, 
 {'name': 'rect', 'x': 140, 'y': 42, 'width': 30, 'height': 45}]

Answer 2

我找到了使用 json 的解决方案

import json

strigDicts = [
    '{""name"":""rect"",""x"":161,""y"":562,""width"":51,""height"":96}',
    '{""name"":""rect"",""x"":151,""y"":542,""width"":32,""height"":69}',
    '{""name"":""rect"",""x"":140,""y"":42,""width"":30,""height"":45}'
    ]

dicts = []
for d in strigDicts:
    dicts.append(json.loads(d.replace('""', '"')))