-
Notifications
You must be signed in to change notification settings - Fork 47
/
retrainGaussian.py
190 lines (154 loc) · 8.3 KB
/
retrainGaussian.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
178
179
180
181
182
183
184
185
186
187
188
189
190
from GaussianNet import gaussianNet
from net_functions import *
import os
os.environ["THEANO_FLAGS"] = "device=gpu2, floatX=float32"
from PIL import Image
import pickle
import time
import json
import MyConfig
class gaussian2(gaussianNet):
def __init__(self):
gaussianNet.__init__(self)
self.trainImgsPath = MyConfig.trainImgPath
self.trainLabelsPath = MyConfig.trainLabelPath
self.imgList = []
self.labelList = []
def checkPath(self, path):
if not os.path.exists(path):
os.makedirs(path)
def loadImgList(self, dataPath, data_ext):
files = [f for f in os.listdir(dataPath) if os.path.isfile(dataPath + f)]
files = [i for i in files if i.endswith('.'+data_ext)]
self.imgList = files
def generateLabelList(self, imgList, img_ext):
labelList = [MyConfig.labelName%f[:-(len(img_ext)+1)] for f in imgList]
self.labelList = labelList
def loadjsonData(self, dataPath, jsonfile):
with open(dataPath + jsonfile) as read_file:
data = json.load(read_file)
return np.array(data)
def load_batch(self, local_training_set_indices, train=True, from_generated=False):
batch_size = len(local_training_set_indices)
rgb_list = []
labels_list = []
for idx in local_training_set_indices:
# rgb = np.asarray(Image.open(self.imgs[fid]))[:, :, 0:3]
rgb = np.asarray( Image.open(self.trainImgsPath+self.imgList[idx]) )[:, :, 0:3]
H, W = np.shape(rgb)[0:2]
rgb_theano = rgb.transpose((2, 0, 1))
rgb_theano = rgb_theano.reshape((1, 3, H, W))
rgb_list.append(rgb_theano)
# if train:
# labels = np.clip(np.loadtxt(self.labels_path % fid), -1000, 1000)
CNN_factor = 4
H_lab, W_lab = H / CNN_factor, W / CNN_factor
# print labels.shape
labels = self.loadjsonData(self.trainLabelsPath, self.labelList[idx])
labels = labels.reshape(H_lab, W_lab, 5)
labels = labels.transpose((2, 0, 1))
labels = labels.reshape(1, 5, H_lab, W_lab)
labels_list.append(labels)
x_in = np.concatenate(rgb_list, axis=0)
y_in = np.concatenate(labels_list, axis=0)
return x_in, y_in
def optimize_gaussians_online(self,all_indices,gaussian_minibatch_size = 4,from_generated = False):
number_of_minibatches = len(all_indices)/gaussian_minibatch_size
self.go_zero_sum_func()
#we are computing the sums that will be used to update the gaussians
for b in range(0,number_of_minibatches):
local_indices = all_indices[b*gaussian_minibatch_size:(b+1)*gaussian_minibatch_size]
#print 'gaussian minibatch',b
x_in,y_in = self.load_batch(local_indices,train = True,from_generated = from_generated)
self.train_sums_func(x_in,y_in)
self.train_gaussians(x_in[0:2], y_in[0:2])
return
def train_parts(self, em_it, training_round = 0):
batch_size = 4
generated_training_set_size = len(self.imgList)
epoch_set_size = min(generated_training_set_size,400)
update_gaussian_every_batch_iters = min(generated_training_set_size,100)
gaussian_fitting_size = min(generated_training_set_size,500)
train_logs_path = MyConfig.log_path
self.checkPath(train_logs_path)
params_bg = pickle.load(open(MyConfig.bgParams_path))
self.mBGsub.setParams(params_bg)
if training_round ==0:
log_filename = 'train_%d.txt' % em_it
self.checkPath(train_logs_path)
f_logs = open(train_logs_path+log_filename, 'w')
f_logs.close()
if em_it == 0:
#initilize params
init_gaussian_params = init_all_gaussian_params(self.n_leaves)
load_gaussian_params_fromshared(self.params_gaussian, init_gaussian_params)
random_reg_params = self.regression_net.get_random_regression_params()
self.regression_net.load_regression_params(random_reg_params)
print 'finished initialization'
else:
#load parameters from previous em iteration
params_regression = pickle.load(
open(MyConfig.net_params_path + 'EM%d_params_regression.pickle' % (em_it - 1)))
self.regression_net.load_regression_params(params_regression)
gaussian_params = pickle.load(
open(MyConfig.net_params_path + 'EM%d_params_gaussian.pickle' % (em_it - 1)))
load_gaussian_params(self.params_gaussian, gaussian_params)
else:
#load parameters from previous training iteration
params_regression = pickle.load(open(MyConfig.net_params_path
+ 'params_regression_%d.pickle' % (training_round-1)))
self.regression_net.load_regression_params(params_regression)
gaussian_params = pickle.load(open(MyConfig.net_params_path
+ 'params_gaussian_%d.pickle' % (training_round-1)))
load_gaussian_params(self.params_gaussian, gaussian_params)
#learning regression parameters
for iterIdx in range(training_round, MyConfig.epochs):
print'epoch %d' %(iterIdx)
generated_training_set_order = np.random.permutation(np.arange(0, generated_training_set_size))
# Train
av_cost = 0
for batch in range(0,epoch_set_size/batch_size):
local_training_set_indices = generated_training_set_order[batch*batch_size:(batch+1)*batch_size]
x_in,y_in = self.load_batch(local_training_set_indices,train = True,from_generated = True)
t_start = time.time()
cost = self.train_decision_func(x_in,y_in)[0]
t_end = time.time()
print 'regression training time %f' % (t_end - t_start)
av_cost+=cost
#Optimise gaussian
local_training_set_indices = generated_training_set_order[batch*batch_size:
batch*batch_size+gaussian_fitting_size]
if batch % update_gaussian_every_batch_iters == update_gaussian_every_batch_iters - 1:
self.optimize_gaussians_online(local_training_set_indices,from_generated=True)
av_cost = av_cost / (epoch_set_size/batch_size)
print 'av_cost = %f' %av_cost
f_logs = open(train_logs_path+log_filename, 'a')
f_logs.write('%f' % (av_cost) + '\n')
f_logs.close()
# Save Params after each two iterations
if iterIdx % 2 == 0:
params_regression = self.regression_net.save_regression_params() # load regression params to the var on left side of =
gaussian_params = save_gaussian_params(self.params_gaussian)
self.checkPath(MyConfig.net_params_path)
with open(MyConfig.net_params_path + 'params_regression%d_%d.pickle'%(em_it, iterIdx), 'wb') as a:
pickle.dump(params_regression, a)
with open(MyConfig.net_params_path + 'params_gaussian%d_%d.pickle'%(em_it, iterIdx), 'wb') as a:
pickle.dump(gaussian_params, a)
# Run small test
#self.run_test(em_it, reload_params=False, name='test_em%d_it%d_' % (em_it, iterIdx))
params_regression = self.regression_net.save_regression_params()
gaussian_params = save_gaussian_params(self.params_gaussian)
self.checkPath(MyConfig.net_params_path)
with open(MyConfig.net_params_path + 'EM%d_params_regression.pickle'%em_it,'wb') as a:
pickle.dump(params_regression,a)
with open(MyConfig.net_params_path + 'EM%d_params_gaussian.pickle'%em_it,'wb') as a:
pickle.dump(gaussian_params,a)
def main(reloadData = False):
gaussianModel = gaussian2()
gaussianModel.loadImgList( gaussianModel.trainImgsPath, MyConfig.imgExt)
gaussianModel.generateLabelList( gaussianModel.imgList, MyConfig.imgExt)
print 'start training'
for em_iter in range(MyConfig.iterations):
gaussianModel.train_parts(em_iter)
if __name__ =="__main__":
main()