-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathstep4_plot_confusion_matrix.py
103 lines (99 loc) · 4.45 KB
/
step4_plot_confusion_matrix.py
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
import librosa
import numpy as np
from Util import DataLoad
from Util.dtw import dtw
from scipy.linalg import norm
import os
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import confusion_matrix
from sklearn.preprocessing import normalize
import pandas as pd
import seaborn as sn
import matplotlib.pyplot as plt
from tensorflow.keras.utils import to_categorical
from models import Transfer_VGG, Resnet, CRNN
# 测试数据集和训练数据集
test_dirname = r'data\test'
train_dirname = r'data\train_after_process'
"""绘制混淆矩阵"""
'''以下部分为画KNN的混淆矩阵'''
temp = np.load('save_model/KNN/data_1.npz')
temp2 = np.load('save_model/KNN/data2_1.npz')
distances = temp.f.arr_0
label = temp2.f.arr_0
classifier = KNeighborsClassifier(n_neighbors=3, metric='euclidean')
# fit可以简单的认为是表格存储
classifier.fit(distances, label)
test_dirs = [f for f in os.listdir(test_dirname)]
train_dirs = [f for f in os.listdir(train_dirname)]
test_real_lable = []
test_pred_lable = []
l = 0
for dir1 in test_dirs:
files1 = os.listdir(test_dirname + '/' + dir1)
for i in range(len(files1)):
y1, sr1 = librosa.load(test_dirname + '/' + dir1 + '/' + files1[i])
mfcc1 = librosa.feature.mfcc(y=y1, sr=sr1)
distanceTest = []
for dir2 in train_dirs:
files2 = os.listdir(train_dirname + '/' + dir2)
for j in range(len(files2)):
y2, sr2 = librosa.load(train_dirname + "/" + dir2 + '/' + files2[j])
mfcc2 = librosa.feature.mfcc(y=y2, sr=sr2)
dist, _, _, _ = dtw(mfcc1.T, mfcc2.T, dist=lambda x, y: norm(x - y, ord=1))
distanceTest.append(dist)
pre = classifier.predict([distanceTest])[0]
test_pred_lable.append(pre)
test_real_lable.append(l)
l += 1
my_confusion_matrix = confusion_matrix(test_real_lable, test_pred_lable)
normalized_confusion_matrix = normalize(my_confusion_matrix, norm='l1')
df_cm = pd.DataFrame(normalized_confusion_matrix, test_dirs, train_dirs)
sn.set(font_scale=0.9)
sn.heatmap(df_cm, annot=True, annot_kws={"size": 8}, cmap="Blues")
plt.xticks(ticks=[0.5, 1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5, 9.5, 10.5, 11.5, 12.5], rotation=45)
plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
plt.title('Confusion Matrix About Knn')
plt.savefig(r".\picture\confusion_matrix_knn_1", dpi=900, bbox_inches='tight')
plt.show()
# '''以下部分为画神经网络的混淆矩阵'''
# labels_dict = np.load(r'save_model\label_dict.npy', allow_pickle=True).item()
# labels = list(labels_dict.values())
# print(labels)
# models = ["transfer_vgg16"]
#
# for i in range(len(models)):
# if models[i] == 'transfer_vgg16' or models[i] == 'Resnet':
# Data_loader = DataLoad(test_dirname, pattern=1, delta_dim=2)
# elif models[i] == "CRNN":
# Data_loader = DataLoad(test_dirname, pattern=1, delta_dim=1)
# train_x, valid_x, train_y, valid_y, label_dict = Data_loader.train_data() # 借用训练数据导入程序，原理一样
# test_x = np.concatenate([train_x, valid_x])
# test_y = np.concatenate([train_y, valid_y])
# num_classes = len(np.unique(test_y))
# test_y = to_categorical(test_y)
# input_dim = test_x.shape[1:]
#
# if models[i] == 'transfer_vgg16':
# model = Transfer_VGG(num_classes, input_dim)
# elif models[i] == 'Resnet':
# model = Resnet(num_classes, input_dim)
# elif models[i] == 'CRNN':
# model = CRNN(num_classes, input_dim)
# model.load_weights(r'.\save_model\\' + models[i] + r'\model_weight')
#
# pred = model.predict(test_x)
# test_y_real = np.argmax(test_y, 1)
# test_y_pred = np.argmax(pred, 1)
# my_confusion_matrix = confusion_matrix(test_y_real, test_y_pred)
# normalized_confusion_matrix = normalize(my_confusion_matrix, norm='l1')
# df_cm = pd.DataFrame(normalized_confusion_matrix, labels, labels)
# sn.set(font_scale=0.9)
# sn.heatmap(df_cm, annot=True, annot_kws={"size": 8}, cmap="Blues")
# plt.xticks(ticks=[0.5, 1.5, 2.5, 3.5, 4.5, 5.5, 6.5, 7.5, 8.5, 9.5, 10.5, 11.5, 12.5], rotation=45)
# plt.rcParams['font.sans-serif'] = ['SimHei']
# plt.rcParams['axes.unicode_minus'] = False
# plt.title('Confusion Matrix About ' + models[i])
# plt.savefig(r".\picture\confusion_matrix_" + models[i] , dpi=900, bbox_inches='tight')
# plt.show()