-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_single.py
275 lines (236 loc) · 10 KB
/
train_single.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
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
# -*- coding: utf-8 -*-
"""
A naive convnet that takes the group's label as each instance's label
"""
from time import time
import os
import h5py
import pandas as pd
import numpy as np
from PIL import Image
import cv2
import matplotlib.pyplot as plt
from keras.models import Sequential, model_from_json
from keras.layers.core import Permute, Dense, Flatten, Dropout
from keras.layers.convolutional import Convolution2D, MaxPooling2D
from keras.callbacks import Callback
# convert numeric labels to binary matrix
def to_bool(s):
return(pd.Series([1L if str(i) in str(s).split(' ') else 0L for i in range(9)]))
# take average prediction of binary matrix and convert into a string of numeric labels
def from_bool(x):
x = np.where(x.mean(axis=0).round())[0]
return(' '.join(str(i) for i in x))
# sample up to 10 pictures from each business_id
def take(series,n=10):
n = min(n,len(series))
return(series.sample(n,random_state=290615))
# image generator + augmentor
def image_generator(i_min,i_max,batch_size,augment=True):
rng = np.random.RandomState(290615)
bi,b_list = 0,rng.permutation(range(i_min/batch_size,i_max/batch_size))
while True:
b = b_list[bi]
b_range = range(b*batch_size,(b+1)*batch_size)
X_batch = np.empty((batch_size,h,w,ch))
Y_batch = np.asarray(XY.ix[b_range,['0','1','2','3','4','5','6','7','8']])
for i_ in b_range:
f_ = 'data/train_photos/' + str(XY.ix[i_,'photo_id']) + '.jpg'
im = Image.open(os.path.realpath(f_))
im_sml = im.resize((w,h))
# scale inputs [-1,+1]
xi = np.asarray(im_sml)/128.-1
if augment:
# flip coords horizontally (but not vertically)
if rng.rand(1)[0] > 0.5:
xi = np.fliplr(xi)
# rescale slightly within a random range
jit = w*0.2
if rng.rand(1)[0] > 0.1:
xl,xr = rng.uniform(0,jit,1),rng.uniform(w-jit,w,1)
yu,yd = rng.uniform(0,jit,1),rng.uniform(h-jit,h,1)
pts1 = np.float32([[xl,yu],[xr,yu],[xl,yd],[xr,yd]])
pts2 = np.float32([[0,0],[w,0],[0,h],[w,h]])
M = cv2.getPerspectiveTransform(pts1,pts2)
xi = cv2.warpPerspective(xi,M,(w,h))
# save back to X_batch
X_batch[i_-b*batch_size,:,:,:] = xi
yield([X_batch],Y_batch)
if bi < len(b_list)-1:
bi += 1
else:
bi,b_list = 0,rng.permutation(range(i_min/batch_size,i_max/batch_size))
def test(model,gen,n_id,threshold=0.5,verbose=True):
"""
Evaluate on test set
Parameters
----------
model: keras model
Model
gen: generator
Generates a batch of images for a business_id in each batch
n_id: int
number of business_ids in test set
"""
t_start = time()
ce_avg,tp,tn,fp,fn = 0.,0.,0.,0.,0.
for i in range(n_id):
X_test,y_test = gen.next()
y_test = y_test.mean()
Y_pred = model.predict(X_test)
y_pred = Y_pred.max(axis=0)
ce = np.mean(- y_test * np.log(y_pred) - (1-y_test) * np.log(1-y_pred))
ce_avg += ce/n_id
y_predr = y_pred.round()
tp += sum((y_test == 1) & (y_predr == 1))
tn += sum((y_test == 0) & (y_predr == 0))
fp += sum((y_test == 0) & (y_predr == 1))
fn += sum((y_test == 1) & (y_predr == 0))
print(i)
prec,recall,acc = tp/(tp+fp+1e-15),tp/(tp+fn+1e-15),(tp+tn)/n_id
F1 = 2*tp/(2*tp+fp+fn)
if verbose:
print('Valid F1 %.3f tp %.3f tn %.3f fp %.3f fn %.3f' % (F1,tp,tn,fp,fn))
print('Took %.1fs' % (time()-t_start))
return(ce_avg,prec,recall,F1,acc,tp,tn,fp,fn)
# callback that loops over validation set generator
class Validator(Callback):
def __init__(self, valid_gen, n_id, print_every_n=np.inf):
self.gen = valid_gen
self.n_id = n_id
self.print_every_n = print_every_n
def on_batch_end(self, epoch, logs={}):
if epoch % self.print_every_n == 0:
print('batch: %i acc: %.3f loss: %.3f size: %i' % \
(logs['batch'],logs['acc'],logs['loss'],logs['size']))
def on_epoch_end(self, epoch, logs={}):
ce,prec,recall,F1,acc,tp,tn,fp,fn = test(model,self.gen,self.n_id)
self.logs[epoch] = {'ce':ce,'prec':prec,'recall':recall,'F1':F1,
'acc':acc,'tp':tp,'tn':tn,'fp':fp,'fn':fn}
# train model
def train():
print('Compiling Model')
t_comp = time()
model = Sequential()
# reorder input to ch, h, w (no sample axis)
model.add(Permute((3,1,2),input_shape=(h,w,ch)))
# add conv layers
model.add(Convolution2D(16,3,3,init='glorot_uniform',activation='relu',
subsample=(1,1)))
# model.add(Convolution2D(16,3,3,init='glorot_uniform',activation='relu',
# subsample=(1,1)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(0.1))
model.add(Convolution2D(32,3,3,init='glorot_uniform',activation='relu',
subsample=(1,1)))
# model.add(Convolution2D(32,3,3,init='glorot_uniform',activation='relu',
# subsample=(1,1)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(0.1))
model.add(Convolution2D(64,3,3,init='glorot_uniform',activation='relu',
subsample=(1,1)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Convolution2D(64,3,3,init='glorot_uniform',activation='relu',
subsample=(1,1)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Flatten())
model.add(Dense(output_dim=500,init='glorot_uniform',activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(output_dim=500,init='glorot_uniform',activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(output_dim=9,init='glorot_uniform',activation='sigmoid'))
model.compile(optimizer='rmsprop', loss='binary_crossentropy',
class_mode='binary')
t_train = time()
print('Took %.1fs' % (t_train-t_comp))
# basic fitting with image generator
train_p,valid_p = 0.8,0.1
i_train = int(nrow*train_p)/batch_size*batch_size
i_valid = int(nrow*(train_p+valid_p))/batch_size*batch_size
train_gen = image_generator(0,i_train,batch_size,augment=True)
valid_gen = image_generator(i_train,i_valid,batch_size,augment=False)
test_gen = image_generator(i_valid,nrow,batch_size,augment=False)
validator = Validator(valid_gen,i_valid-i_train,100)
hist_gen = model.fit_generator(train_gen,
samples_per_epoch=i_train,
nb_epoch=100,
verbose=2,
show_accuracy=True,
callbacks=[validator])
# evaluate on test set
ce,prec,recall,F1,acc,tp,tn,fp,fn = test(model,test_gen,nrow-i_valid)
# save model
mname = 'model_%03d' % (loss_test*100)
model_json = model.to_json()
open('models/%s.json' % (mname), 'w').write(model_json)
model.save_weights('models/%s_weights.h5' % (mname))
return(loss_test)
# visualize selected/flagged examples on random data
# TODO adapt to multi-label at once
def show_classified_pics(gen,model):
X_batch,y_batch = gen.next()
y_pred = model.predict(X_batch[0])
y_predr = np.max(y_pred.round())
plt.rcParams['figure.figsize'] = (14,14)
xis = np.where(y_pred > 0.5)
fig1 = plt.figure()
fig1.suptitle('Label 1',fontsize=24)
for i in range(len(xis[0])):
axi = fig1.add_subplot(6,6,i+1)
i_ = xis[0][i]
axi.imshow((X_batch[0][i_,...]+1)/2)
axi.axis('off')
axi.set_title('%.3f' % y_pred[i_])
xis = np.where(y_pred < 0.5)
fig2 = plt.figure()
fig2.suptitle('Label 0',fontsize=24)
for i in range(len(xis[0])):
axi = fig2.add_subplot(6,6,i+1)
i_ = xis[0][i]
axi.imshow((X_batch[0][i_,...]+1)/2)
axi.axis('off')
axi.set_title('%.3f' % y_pred[i_])
print('Truth %i \nPredicted %i' % (y_batch[0],y_predr))
return(X_batch,y_batch)
# store test images as a matrix for making predictions
def store_test():
data = h5py.File('data/test_matrix.h5','w')
X_test = data.create_dataset(
name='X',
shape=(len(XY_test),h,w,ch),
dtype=np.float32)
for i in range(len(XY_test)):
f_ = 'data/test_photos/%s.jpg' % (XY_test['photo_id'][i])
im_sml = Image.open(f_).resize((w,h))
X_test[i,:,:,:] = np.asarray(im_sml)/128.-1
if i % 1000 == 0:
print('reading %ith image' % i)
data.close()
if __name__ == '__main__':
# prep data to be read
train = pd.read_csv('data/train.csv')
biz_id_train = pd.read_csv('data/train_photo_to_biz_ids.csv')
train[['0','1','2','3','4','5','6','7','8']] = train['labels'].apply(to_bool)
XY = pd.merge(biz_id_train,train,on='business_id')
h,w,ch,batch_size = 128,128,3,32
nrow,ncol,ny = len(XY),h*w*ch,9
loss_test = train()
# predict
model = model_from_json(open('models/model_%03d.json' % (loss_test*100)).read())
model.load_weights('models/model_%03d_weights.h5' % (loss_test*100))
biz_id_test = pd.read_csv('data/test_photo_to_biz.csv')
submit = pd.read_csv('data/sample_submission.csv')
# take average of up to 10 photos for each biz_id
test_id = biz_id_test.groupby('business_id').apply(take)
XY_test = pd.merge(test_id,submit,on='business_id')
# store_test()
data = h5py.File('data/test_matrix.h5','r')
X_test = data['X']
Y_pred = pd.DataFrame(model.predict(X_test,batch_size=128)) # 46mins
data.close()
Y_pred['business_id'] = XY_test['business_id']
labels = pd.DataFrame(Y_pred.groupby('business_id').apply(from_bool))
labels['business_id'],labels['labels'] = labels.index,labels[0]
submit_labels = pd.merge(submit[['business_id']],labels[['business_id','labels']])
submit_labels.to_csv('data/sub2_singleconv.csv',index=False)
# mostly 1 2 5 6 8. F1 score of ~0.65