Converting a nested JSON array to pandas dataframe

You can parse the string as JSON using json.loads():

import json

# defining the provided string
dat =  '{"ID":"rtbg345h","DataRow ID":"dgdfg45654","Labeled Data":"https://abc.png","Label":{"objects":[{"featureId":"rtbg345h","schemaId":"rtbg345h","title":"iris","value":"flower","color":"#00RGAA","dpoin":{"top":82,"left":33,"height":52,"width":675},"instanceURI":"https://sdfdsf.ab"}],"classifications":[]},"Created By":"user","Project Name":"myfirstproject","Created At":"2018-02-02","Updated At":"2018-02-02","Seconds to Label":24.264,"External ID":"sds.jpg","Agreement":-1,"Benchmark Agreement":-1,"Benchmark ID":null,"Dataset Name":"mine","Reviews":[],"View Label":"https:fdrtdf"}'

# parsing string as JSON
res = json.loads(dat) 

Here's what the parsed JSON looks like:

{'ID': 'rtbg345h',
 'DataRow ID': 'dgdfg45654',
 'Labeled Data': 'https://abc.png',
 'Label': {'objects': [{'featureId': 'rtbg345h',
    'schemaId': 'rtbg345h',
    'title': 'iris',
    'value': 'flower',
    'color': '#00RGAA',
    'dpoin': {'top': 82, 'left': 33, 'height': 52, 'width': 675},
    'instanceURI': 'https://sdfdsf.ab'}],
  'classifications': []},
 'Created By': 'user',
 'Project Name': 'myfirstproject',
 'Created At': '2018-02-02',
 'Updated At': '2018-02-02',
 'Seconds to Label': 24.264,
 'External ID': 'sds.jpg',
 'Agreement': -1,
 'Benchmark Agreement': -1,
 'Benchmark ID': None,
 'Dataset Name': 'mine',
 'Reviews': [],
 'View Label': 'https:fdrtdf'}

Then, access the elements using the appropriate notation (note that res['Label']['objects'] is a list containing a single dictionary)

title = res['Label']['objects'][0]['title']
dpoin = res['Label']['objects'][0]['dpoin']

If you have a list of JSON objects, you can parse each one individually, storing the parsed data in a list. Once you've parsed all the JSON objects, you can create a dataframe from the list.

Here's an example that keeps track of the data in a list of dictionaries:

l = [dat, dat] # repeating the obj you provided to make an example list

rows = []

for item in l: 
    res = json.loads(item)
    row = res['Label']['objects'][0]['dpoin']
    row['title'] = res['Label']['objects'][0]['title']
    rows.append(row)

df = pd.DataFrame.from_dict(rows)
df

Resulting in this dataframe:

   top  left  height  width title
0   82    33      52    675  iris
1   82    33      52    675  iris

This example creates a dictionary row for each JSON object, contaning all the fields from dpoin as well as the title. Storing the data as a list of dictionaries allows you to use the pd.DataFrame.from_dict() constructor.