Reading data from a json using Python and using it to draw shapes

Question

I have a COCO formatted json which if you are not familiar with COCO, is just a list of coordinates of 10 sided polygons (attached an example).

I need to create an array from these coordinates that will plot them on a 128 x 128 square as filled in polygons so that I can multiply it with another array.

I have so far tried skimage.draw polygons and matplotlib polygons. These work if I manually enter coordinates, but I don't know how to read the data in from the json and then use that as the coordinates.

Below is an example of the json.

{
"version": "4.6.0",
"flags": {},
"shapes": [
{
  "label": "blob",
  "points": [
    [
      75.14285714285714,
      83.71428571428571
    ],
    [
      72.23121416791264,
      82.67540136440437
    ],
    [
      70.48628641932781,
      80.12350546954684
    ],
    [
      70.57457698914924,
      77.03333552579737
    ],
    [
      72.46236188059122,
      74.58523142065489
    ],
    [
      75.42857142857144,
      73.71428571428571
    ],
    [
      78.34021440351594,
      74.75317006416705
    ],
    [
      80.08514215210077,
      77.30506595902457
    ],
    [
      79.99685158227935,
      80.39523590277405
    ],
    [
      78.10906669083737,
      82.84334000791652
    ]
  ],
  "group_id": null,
  "shape_type": "polygon",
  "flags": {}

I have tried:

f=open('139cm_2000_frame27.json') 
data=json.load(f) 
shapes=data["shapes"] 
for i in shapes: 
    print('points')

but it errors at 'points' Thank you so much!

Answer 1

You can use the json library to work with json objects. If you need to load the json from a file use json.load() like so:

import json
with open('example.json', 'r') as f:
    versions = json.load(f)['versions']

and use loads when from a string:

import json
json_str = '{"versions": 10}'
versions = json.loads(json_str)['versions']

Answer 2

I would have chosen jmespath libray for this use case to extract points from the json pip install jmespath just check it out

#we are searching in shapes list and fetching out label and points from it it will give us list of dictionary for each shapes
import jmespath
expression = jmespath.compile('shapes[*].{Name: label, points: points}')
for shape_name,points in expression.search(data)[0].items():
    print(shape_name, points)

Answer 3

Ok, you’ve managed to get some way. You now have to navigate the ‘dicts within lists within dicts’. This code will simply print out all the coordinates of all the shapes:

f=open('139cm_2000_frame27.json') 
data=json.load(f) 
shapes=data["shapes"] 
for i in shapes: 
    print(i['label'])   # prints the label first
    for c in i['points']:
        print(c)       # a list containing coordinates