CoralNet-API/coralnet_api_utils.py at master · JordanMakesMaps/CoralNet-API · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
import json
import numpy as np
import pandas as pd

def get_points(height, width, h_offset, w_offset, percentage, sampling_method):
    '''
    height: the height of the image (rows)
    width: the width of the image (columns)
    offset: the % of pixels on all sides to avoid sampling (i.e. avoids edges of image)
    percentage: the % of points to sample (1% of a 4MP image = 40,000 points)
    sampling_method: either "random" or "grid"
    '''

    percentage = percentage * .01

    num_points = int(height * width * percentage)

    if(sampling_method == 'random'):

        x = np.random.randint(w_offset, width - w_offset, num_points)
        y = np.random.randint(h_offset, height - h_offset, num_points)

    else:

        density = int(np.sqrt(num_points))

        # Creates an equally spaced grid, reshapes, converts into list
        x_, y_ = np.meshgrid(np.linspace(w_offset, width - w_offset, density),
                             np.linspace(h_offset, height - h_offset, density))

        xy = np.dstack([x_, y_]).reshape(-1, 2).astype(int)

        x = [point[0] for point in xy]
        y = [point[1] for point in xy]

        # Any labels that did not fit in the grid will be sampled randomly
        x += np.random.randint(w_offset, width - w_offset, num_points - len(xy)).tolist()
        y += np.random.randint(h_offset, height - h_offset, num_points - len(xy)).tolist()


    points = []

    for _ in range(num_points):
        points.append({'row': int(y[_]), 'column': int(x[_])})

    return points


def decode_status(r_status):

    curr_status = json.loads(r_status.content)
    message = ''

    if 'status' in curr_status['data'][0]['attributes'].keys():

        s = curr_status['data'][0]['attributes']['successes']
        f = curr_status['data'][0]['attributes']['failures']
        t = curr_status['data'][0]['attributes']['total']
        status = curr_status['data'][0]['attributes']['status']
        ids = curr_status['data'][0]['id'].split(",")
        ids = ''.join(str(_) for _ in ids)

        message = 'Success: ' + str(s) + ' Failures: ' + str(f) + ' Total: ' + str(t) + ' Status: ' + str(status) + ' Ids: ' + ids

    return curr_status, message


def check_status(r):

    # Sends a request to retrieve the completed annotations, obtains status update
    r_status = requests.get(url = 'https://coralnet.ucsd.edu' + r.headers['Location'],
                            headers = {"Authorization": f"Token {coralnet_token}"})

    # Extracts the content from the status update
    curr_status, message = decode_status(r_status)

    return curr_status, message


def convert_to_csv(export):

    all_preds = pd.DataFrame()

    image_file = export['data'][0]['id'].split("/")[-1].split("?")[0]

    for point in export['data'][0]['attributes']['points']:

        per_point = dict()

        per_point['image'] = image_file

        per_point['X'] = point['column']
        per_point['Y'] = point['row']

        for index, classification in enumerate(point['classifications']):

            per_point['score_' + str(index + 1)] = classification['score']
            per_point['label_id_' + str(index + 1)] = classification['label_id']
            per_point['label_code_' + str(index + 1)] = classification['label_code']
            per_point['label_name_' + str(index + 1)] = classification['label_name']

        all_preds = pd.concat([all_preds, pd.DataFrame.from_dict([per_point])])

    return all_preds