-
Notifications
You must be signed in to change notification settings - Fork 1
/
AVE_modify.py
126 lines (103 loc) · 4.59 KB
/
AVE_modify.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
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
import os
import moviepy.editor as mp
import cv2
import h5py
from scipy import signal
import soundfile as sf
import resampy
import numpy as np
from utils.DataFromtxt import readDataTxt
Data_path = r"/home/ubuntu/AVE_Dataset"
Video_path = r"/home/ubuntu/AVE_Dataset/Video"
Pic_path = r"/media/ubuntu/Data/Picture"
Audio_path = r"/media/ubuntu/Data/Audio"
H5_path = r"/media/ubuntu/Data/H5"
def save_image(image, addr, num, point):
if point == 2:
address = os.path.join(addr, "{:0>2d}".format(num) + '.jpg')
else:
address = os.path.join(addr, "{:0>4d}".format(num) + '.jpg')
cv2.imwrite(address, image)
def mp4_to_jpg(video_path, pic_save_path, start_time=0, end_time=10, point=2):
_, video_name = os.path.split(video_path)
folder_name = video_name.split('.')[0] # mp4文件名，无后缀,也是目标地址的文件夹名
folder_path = os.path.join(pic_save_path, folder_name) # 新的目录存放每个视频的图片
os.makedirs(folder_path, exist_ok=True) # 创建存放视频的对应目录
video = cv2.VideoCapture(video_path) # 读入视频文件
if not video.isOpened():
return
fps = video.get(cv2.CAP_PROP_FPS) # 获取帧率
print(fps) # 帧率可能不是整数 需要取整
rval, frame = video.read() # videoCapture.read() 函数，第一个返回值为是否成功获取视频帧，第二个返回值为返回的视频帧
fps_id = 1
while rval: # 循环读取视频帧
if fps_id % round(fps) == round(fps) // 2 and start_time <= fps_id // round(fps) <= end_time: # 每隔fps帧进行存储操作 ,可自行指定间隔
save_image(frame, folder_path, fps_id // round(fps), point)
# cv2.cvWaitKey(1) # waitKey()--这个函数是在一个给定的时间内(单位ms)等待用户按键触发;如果用户没有按下 键,则接续等待(循环)
rval, frame = video.read()
fps_id = fps_id + 1
video.release()
print('save_success ' + folder_path)
def generate_jpg(mode):
data_list = readDataTxt(Data_path, mode)
for data in data_list:
mp4_to_jpg(os.path.join(Video_path, data[0] + ".mp4"), Pic_path) # , data[-2], data[-1])
def mp4_to_wav(videos_path, to_path, start_time=0, end_time=10):
if not os.path.exists(to_path):
os.makedirs(to_path)
my_clip = mp.VideoFileClip(videos_path)
if end_time - start_time != 10:
my_clip.subclip(start_time, end_time)
_, videos_name = os.path.split(videos_path)
audio_path = os.path.join(to_path, videos_name[:-4]+".wav")
my_clip.audio.write_audiofile(audio_path)
def generate_wav(mode):
if not os.path.exists(Audio_path):
os.makedirs(Audio_path)
data_list = readDataTxt(Data_path, mode)
for data in data_list:
mp4_to_wav(os.path.join(Video_path, data[0]+ ".mp4"), Audio_path) # data[-2], data[-1]
def wav_to_h5(audio_path, to_path, point=2):
Audio_file_path = audio_path + ".wav"
_, Audio_name = os.path.split(audio_path)
os.makedirs(os.path.join(to_path, Audio_name), exist_ok=True)
samples, samplerate = sf.read(Audio_file_path)
if len(samples.shape) > 1:
samples = np.mean(samples, axis=1)
SAMPLE_RATE = 16000
if samplerate != SAMPLE_RATE:
samples = resampy.resample(samples, samplerate, SAMPLE_RATE) # 采样速率转换44100->16000
num = samples.size // SAMPLE_RATE
for picp in range(0, num):
s = picp * 16000
e = (picp + 1) * 16000
resamples = samples[s:e]
resamples[resamples > 1.] = 1.
resamples[resamples < -1.] = -1.
_, _, spectrogram = signal.spectrogram(resamples, SAMPLE_RATE, nperseg=160, noverlap=80)
spectrogram = np.log(spectrogram + 1e-7)
mean = np.mean(spectrogram)
std = np.std(spectrogram)
audio_output = np.divide(spectrogram - mean, std + 1e-9) # 257，61
if point == 2:
h5_path = os.path.join(to_path, Audio_name, "{:0>2d}".format(picp) + ".h5")
else:
h5_path = os.path.join(to_path, Audio_name, "{:0>4d}".format(picp) + ".h5")
with h5py.File(h5_path, 'w') as hf:
hf.create_dataset("dataset", data=audio_output)
print(Audio_name, ".h5 is ok")
def generate_h5(mode='train'):
if not os.path.exists(H5_path):
os.makedirs(H5_path)
data_list = readDataTxt(Data_path, mode)
for data in data_list:
wav_to_h5(os.path.join(Audio_path, data[0]), H5_path)
def modefy_data():
generate_jpg("train")
generate_jpg("test")
generate_wav("train")
generate_wav("test")
generate_h5("train")
generate_h5("test")
if __name__ == "__main__":
modefy_data()