forked from openbci-archive/OpenBCI_Python
-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathloader.py
177 lines (158 loc) · 8.4 KB
/
loader.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
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
import os
import matplotlib
from utils import data_types_utils
from utils.dataset_writer_utils import read_and_decode, create_sample_from_image
from utils.utils import get_label
matplotlib.use('Agg')
import matplotlib.pyplot as plt
plt.interactive(False)
import numpy as np
import tensorflow as tf
import time
import librosa
import librosa.display
import numpy
import json
from PIL import Image
from preprocessing.processor import Clip
class DataLoader:
def __init__(self, project_dir, dataset_dir):
meta_info_file = project_dir + "/config/config.json"
with open(meta_info_file) as data_file:
meta_info = json.load(data_file)
self.conf = meta_info
# TODO separate out common stuff
meta_info = meta_info["processing"]["train"]
self.batch_process_threads_num = int(meta_info["batch_process_threads_num"])
self.project_dir = project_dir
self.dataset_dir = dataset_dir
self.num_epochs = int(meta_info["num_epochs"])
self.batch_size = int(meta_info["batch_size"])
self.train_dir = project_dir + str(meta_info["dir"])
self.train_file = str(meta_info["data_set_name"])
self.tfrecords_filename = project_dir + str(meta_info["tfrecords_filename"])
self.number_of_class = int(meta_info["number_of_class"])
self.generated_image_width = int(meta_info["generated_image_width"])
self.generated_image_height = int(meta_info["generated_image_height"])
self.feature_vector_size = int(self.conf["feature_vector_size"])
self.num_channels = int(self.conf["number_of_channels"])
self.conf["processing"]["train"]["number_of_channels"] = self.num_channels
self.generated_image_dir = project_dir + str(meta_info["generated_image_dir"])
self.sampling_rate = self.conf["sampling_rate"]
# TODO add validate and test initialization
def get_train_config(self):
return self.conf["processing"]["train"]
# def save_plot_clip_overview(self, clip, i):
# with clip.audio as audio:
# figure = plt.figure(figsize=(self.generated_image_width, self.generated_image_height), dpi=1)
# axis = figure.add_subplot(111)
# plt.axis('off')
# plt.tick_params(axis='both', left='off', top='off', right='off', bottom='off', labelleft='off',
# labeltop='off',
# labelright='off', labelbottom='off')
# result = np.array(np.array(clip.feature_list['fft'].get_logamplitude()[0:1]))
# librosa.display.specshow(result, sr=clip.RATE, x_axis='time', y_axis='mel', cmap='RdBu_r')
# extent = axis.get_window_extent().transformed(figure.dpi_scale_trans.inverted())
# clip.filename = self.generated_image_dir + clip.filename + str(i) + str("_.jpg")
# plt.savefig(clip.filename, format='jpg', bbox_inches=extent, pad_inches=0, dpi=1)
# plt.close()
# return clip.filename
# def save_clip_overview(self, categories=5, clips_shown=1, clips=None):
# for c in range(0, categories):
# for i in range(0, clips_shown):
# self.save_plot_clip_overview(clips[c][i], i)
# def create_one_big_file(self, file_type):
# writer = tf.python_io.TFRecordWriter(self.tfrecords_filename)
# for directory in sorted(os.listdir('{0}/'.format(self.dataset_dir))):
# store_location = self.generated_image_dir + directory
# # todo make directory if not created
# directory = '{0}/{1}'.format(self.dataset_dir, directory)
# if os.path.isdir(directory) and os.path.basename(directory)[0:3].isdigit():
# print('Parsing ' + directory)
# for clip in sorted(os.listdir(directory)):
# if clip[-3:] == file_type:
# clip_label, clip_data, rows, _ = self.extracted_sample(directory, clip, file_type)
# for j in range(0, rows - 2):
# clip_filename = self.draw_sample_plot_and_save(clip_data, store_location, clip, j)
# sample = create_sample_from_image(clip_filename, clip_label, self.get_train_config())
# writer.write(sample.SerializeToString())
# writer.close()
# return
# print('All {0} recordings loaded.'.format(self.train_file))
def extracted_sample(self, directory, clip, file_type):
print ('{0}/{1}'.format(directory, clip))
clip_category = ('{0}/{1}'.format(directory, clip), directory.split("/0")[1].
split("-")[0].strip())[1]
clip_data = Clip('{0}/{1}'.format(directory, clip), file_type). \
get_feature_vector()
rows = clip_data.shape[0]
cols = clip_data.shape[1]
clip_label = get_label(int(clip_category), self.number_of_class).tostring()
return clip_label, clip_data, rows, cols
# def draw_sample_plot_and_save(self, raw_data_clip, store_location, clip, index):
# figure = plt.figure(figsize=(
# np.ceil(self.generated_image_width + self.generated_image_width * 0.2),
# np.ceil(self.generated_image_height + self.generated_image_height * 0.2)), dpi=1)
# axis = figure.add_subplot(111)
# plt.axis('off')
# plt.tick_params(axis='both', left='off', top='off', right='off', bottom='off',
# labelleft='off',
# labeltop='off',
# labelright='off', labelbottom='off')
# result = np.array(np.array(raw_data_clip[index:index + 1]))
# librosa.display.specshow(result, sr=self.sampling_rate, x_axis='time', y_axis='mel', cmap='RdBu_r')
# extent = axis.get_window_extent().transformed(figure.dpi_scale_trans.inverted())
# clip_filename = "%s%s%s%s" % (store_location, clip, str(index), "_.jpg")
# plt.savefig(clip_filename, format='jpg', bbox_inches=extent, pad_inches=0)
# plt.close(figure)
# return clip_filename
def inputs(self):
with tf.name_scope('input'):
filename_queue = tf.train.string_input_producer([self.tfrecords_filename],
num_epochs=self.num_epochs)
image, label = read_and_decode(filename_queue, self.conf)
images, sparse_labels = tf.train.shuffle_batch(
[image, label], batch_size=self.batch_size, num_threads=self.batch_process_threads_num,
capacity=1000 + 3 * self.batch_size,
min_after_dequeue=100)
return images, sparse_labels
def run_training(self):
with tf.Graph().as_default():
image, label = self.inputs()
with tf.Session() as sess:
init_op = tf.group(tf.global_variables_initializer(),
tf.local_variables_initializer())
sess.run(init_op)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)
try:
step = 0
while not coord.should_stop():
start_time = time.time()
while not coord.should_stop():
# Run training steps or whatever
exatmple, l = sess.run([image, label])
print exatmple
except tf.errors.OutOfRangeError:
print('Done training for %d epochs, %d steps.' % (self.num_epochs, self.batch_size))
finally:
coord.request_stop()
coord.join(threads)
sess.close()
# parser = argparse.ArgumentParser(description='Data set location and project location')
# parser.add_argument('-dataset_dir', nargs=2)
# parser.add_argument('-project_dir', nargs=1)
#
# opts = parser.parse_args()
#
# project_dir = opts.project_dir
# dataset_dir = opts.dataset_dir
# project_dir = "/home/runge/openbci/git/OpenBCI_Python"
# dataset_dir = "/home/runge/openbci/git/OpenBCI_Python/build/dataset"
#
# loader = DataLoader(project_dir, dataset_dir)
# # # # clips_10 = loader.load_dataset_from_ogg('/home/runge/projects/sound_detector/TRAIN-10')
# # loader.create_one_big_file("ogg")
#
# image, label = loader.inputs()
# loader.run_training()