-
Notifications
You must be signed in to change notification settings - Fork 2
/
PlanetNEt_cnn_predict.py
218 lines (160 loc) · 6.39 KB
/
PlanetNEt_cnn_predict.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
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import tensorflow as tf
from matplotlib import cm
import matplotlib.gridspec as gridspec
import pylab as pl
from classes.PlanetNet_cnn import PlanetNet_cnn
tf.logging.set_verbosity(tf.logging.INFO)
####################################################
#NOTES
'''
working:
model_path='./tmp/saturn_1010_ktest4/storm1010',
# model_path='./tmp/saturn_set2test/storm1010',
'''
####################################################
# train_data = np.load('vims_map_train.npy')
# train_data = np.transpose(np.asarray(train_data,dtype=np.float32))
# train_labels = np.load('vims_labels_train.npy')
# train_labels = np.asarray(train_labels,dtype=np.int32)
# eval_data = np.load('vims_map_test.npy')
# eval_data = np.transpose(np.asarray(eval_data,dtype=np.float32))
# eval_labels = np.load('vims_labels_test.npy')
# eval_labels = np.asarray(eval_labels,dtype=np.int32)
data_prefix = 'data/storm_1010/10-10_saturn' #GOOD
# data_prefix = 'data/storm_set2test/10-10_saturn_set2train_test'
# data_prefix = 'data/storm_set3/20-20_saturn'
# data_prefix = 'data/storm_pca0/20-20_saturn_pca'
# data_prefix = 'data/storm_giant/10-10_saturn_giant_s/torm'
# save_suffix = 'storm_1010_giant_ktest4'
# save_suffix = 'storm_ktest5_1' #GOOD
save_suffix = 'storm_test'
#
map_data = np.load('{}_map_full_resampled_maponly.npy'.format(data_prefix))
map_data = np.transpose(np.asarray(map_data,dtype=np.float32))
spec_data = np.load('{}_spec_full_resampled_maponly.npy'.format(data_prefix))
spec_data = np.transpose(np.asarray(spec_data,dtype=np.float32))
labels = np.load('{}_labels_full_resampled_maponly.npy'.format(data_prefix))
# map_data = np.load('{}_map_full_resampled_speconly.npy'.format(data_prefix))
# map_data = np.transpose(np.asarray(map_data,dtype=np.float32))
# spec_data = np.load('{}_spec_full_resampled_speconly.npy'.format(data_prefix))
# spec_data = np.transpose(np.asarray(spec_data,dtype=np.float32))
# labels = np.load('{}_labels_full_resampled_speconly.npy'.format(data_prefix))
# map_data = np.load('{}_map_full_resampled.npy'.format(data_prefix))
# map_data = np.transpose(np.asarray(map_data,dtype=np.float32))
# spec_data = np.load('{}_spec_full_resampled.npy'.format(data_prefix))
# spec_data = np.transpose(np.asarray(spec_data,dtype=np.float32))
# labels = np.load('{}_labels_full_resampled.npy'.format(data_prefix))
#
# map_data = np.load('{}_map_full.npy'.format(data_prefix))
# map_data = np.transpose(np.asarray(map_data,dtype=np.float32))
# spec_data = np.load('{}_spec_full.npy'.format(data_prefix))
# spec_data = np.transpose(np.asarray(spec_data,dtype=np.float32))
# labels = np.load('{}_labels_full.npy'.format(data_prefix))
# train_labels = np.asarray(train_labels,dtype=np.int32)
# #initialise PlanetNet
# robcnn = PlanetNet_cnn(Nx=12,Ny=12,
# model_path='./tmp/test.ckpt',
# model_type='spectral')
#getting predictions
# predictions = robcnn.predict_single(vims_data)
# predictions = robcnn.predict_single(spec_data)
#convert to one hot labels
Ntrain = len(labels)
Nmax = np.max(labels)+1
Nlabel = 5
print('Map data: ',np.shape(map_data))
print('Spec data: ',np.shape(spec_data))
#initialise PlanetNet
robcnn = PlanetNet_cnn(Nx=10,Ny=10,Ns=256,
Nlogit=Nlabel,
model_path='./tmp/saturn_1010_test2018_20percent/storm1010',
# model_path='./tmp/saturn_1010_ktest4/storm1010',
# model_path='./tmp/saturn_set2test/storm1010',
model_type='both')
#initialise PlanetNet
# robcnn = PlanetNet_cnn(Nx=20,Ny=20,Ns=256,
# Nlogit=5,
# model_path='./tmp/storm_2/storm2',
# model_type='spatial')
#getting predictions
predictions = robcnn.predict_both(map_data,spec_data)
# predictions = robcnn.predict_single(spec_data)
# predictions = robcnn.predict_single(map_data)
cnn_layers = robcnn.get_cnn_filters(plot=False)
#getting prediction classes
pred_labels_flat = predictions['classes']
pred_prob = np.asarray(predictions['probabilities']).reshape(64,64,5)
# pred_prob_dict[i] = np.asarray(pred_prob[:,i]).reshape(64,64)
orig_labs = labels.reshape(64,64)
pred_labs = np.asarray(pred_labels_flat).reshape(64,64)
np.save('PlanetNet_labels_pred_{0}.npy'.format(save_suffix),pred_labs)
np.save('PlanetNet_labels_pred_prob_{0}.npy'.format(save_suffix),pred_prob)
cmap = cm.get_cmap('Set1')
# pl.figure()
# pl.imshow(orig_labs,cmap=cmap,vmin=0,vmax=9)
# pl.title('Original labels')
#
#
# pl.figure()
# pl.imshow(pred_labs,cmap=cmap,vmin=0,vmax=9)
# pl.title('Predicted labels')
#
#
# pl.figure()
# ax = pl.imshow(pred_prob[:,:,0],cmap=cm.get_cmap('viridis'))
# pl.colorbar(ax)
# pl.title('Probability Label: {}'.format(0))
#
# pl.figure()
# ax = pl.imshow(pred_prob[:,:,1],cmap=cm.get_cmap('viridis'))
# pl.colorbar(ax)
# pl.title('Probability Label: {}'.format(1))
#
#
# pl.figure()
# ax = pl.imshow(pred_prob[:,:,2],cmap=cm.get_cmap('viridis'))
# pl.colorbar(ax)
# pl.title('Probability Label: {}'.format(2))
#
# pl.figure()
# ax = pl.imshow(pred_prob[:,:,3],cmap=cm.get_cmap('viridis'))
# pl.colorbar(ax)
# pl.title('Probability Label: {}'.format(3))
#
# pl.figure()
# ax = pl.imshow(pred_prob[:,:,4],cmap=cm.get_cmap('viridis'))
# pl.colorbar(ax)
# pl.title('Probability Label: {}'.format(4))
#ADD prediction comparison plots for paper
pred_errors_flat = np.zeros_like(labels)
pred_errors_flat[labels==pred_labels_flat] = 1.0
pred_errors_sum = np.sum(pred_errors_flat)
pred_errors_frac = pred_errors_sum / len(pred_errors_flat)
print('Error fraction: ',pred_errors_frac)
pred_errors = np.reshape(pred_errors_flat,(64,64))
pred_errors = np.ma.masked_where(pred_errors > 0.9, pred_errors)
fig = pl.figure(figsize=(15,6))
gs = gridspec.GridSpec(1,3)
gs.update(wspace=0.025, hspace=0.25)
ax1 = pl.subplot(gs[0])
ax2 = pl.subplot(gs[1])
ax3 = pl.subplot(gs[2])
ax1.set_yticks([])
ax1.set_xticks([])
ax2.set_yticks([])
ax2.set_xticks([])
ax3.set_yticks([])
ax3.set_xticks([])
# pl.figure()
ax1.set_title('Original labels',fontsize=20)
ax1.imshow(orig_labs,cmap=cmap,vmin=0,vmax=9)
ax2.set_title('Predicted labels',fontsize=20)
ax2.imshow(pred_labs,cmap=cmap,vmin=0,vmax=9)
ax3.set_title('Prediction errors',fontsize=20)
# ax3.imshow(orig_labs,cmap=cmap,vmin=0,vmax=9)
ax3.imshow(pred_errors,alpha=1.0,cmap=cm.get_cmap('binary_r'),vmin=0,vmax=1)
pl.show()