CVPR2022-Biometrics-Workshop-Pet-Biometric-Challenge/ensemble_by_sort.py at main · kali20gakki/CVPR2022-Biometrics-Workshop-Pet-Biometric-Challenge · 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
105
106
107
from glob import glob
import pandas as pd
import os, cv2
from PIL import Image
import numpy as np
from prettytable import PrettyTable
from copy import deepcopy


data = pd.read_csv('dataset/pet_biometric_challenge_2022/test/test_data.csv', encoding='utf-8')

images_pair = [(imageA_name, imageB_name) for imageA_name, imageB_name in zip(data['imageA'], data['imageB'])]


def get_res(file_path):
    data = pd.read_csv(file_path, encoding='utf-8')
    res = [pred_score for i, pred_score in enumerate(data['prediction'])]

    return res


def get_sortid(x):
    index = np.argsort(x)
    out = np.zeros_like(index)
    for i, idx in enumerate(index):
        out[idx] = i

    return np.array([out])


def ensemble_by_sort(res_list):
    out_list = []
    ensemble_res = []
    for res in res_list:
        out_list.append(get_sortid(res))

    out_list_ = np.array(deepcopy(out_list))

    all_out = np.concatenate(out_list, 0)

    mean_out = list(np.mean(all_out, 0))
    x_min, x_max = np.min(mean_out), np.max(mean_out)

    for x in mean_out:
        ensemble_res.append((x - x_min) / (x_max - x_min))

    out_list_ = np.squeeze(out_list_, 1)
    out_list_ = [list(out) for out in out_list_]
    return ensemble_res, out_list_, list(get_sortid(ensemble_res)[0])


def ensemble(file_path_list, show_len = -1):
    res_list = []
    file_names = []
    for file_path in file_path_list:
        res_list.append(get_res(file_path))
        file_names.append(file_path.split('\\')[-1])

    table = PrettyTable()

    merge_res, out_list, merge_index = ensemble_by_sort(res_list)

    res_and_index_list = []
    merge_res_and_index = [[res, i] for res, i in zip(merge_res, merge_index)]

    for res, index in zip(res_list, out_list):
        temp = []
        for r, i in zip(res, index):
            temp.append([r, i])
        res_and_index_list.append(temp)

    for name, res in zip(file_names, res_and_index_list):
        table.add_column(name, res[:show_len])
    table.add_column('ensemble res', merge_res_and_index[:show_len])
    #print(table)
    # with open('merge_table.txt', 'w') as f:
    #     f.write(str(table))

    return merge_res


if __name__ == '__main__':
    # 改为所有结果保存的路径
    all_results_dir = 'test_results'


    file_path_list = [p for p in glob(all_results_dir + '/*.csv')]
    merge_res = ensemble(file_path_list)
    res = []
    imageA_list = []
    imageB_list = []
    for (imageA, imageB), pred in zip(images_pair, merge_res):
        imageA_list.append(imageA)
        imageB_list.append(imageB)
        res.append(pred)


    d = {'imageA': imageA_list, 'imageB': imageB_list, 'prediction': res}

    df = pd.DataFrame(d)

    # 保存 dataframe
    df.to_csv(f'./final_submmit_{len(file_path_list)}res.csv', index=None)
    print(f'final_submmit_{len(file_path_list)}res.csv')