-
Notifications
You must be signed in to change notification settings - Fork 0
/
scratch.py
272 lines (201 loc) · 8.15 KB
/
scratch.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
import numpy as np
import os;
import util;
import random;
import scipy.misc;
import cPickle as pickle;
import visualize;
import processOutput as po;
import scipy.io;
import scipy.stats;
import cv2
def getTifsFromTrain(old_train_txt):
tif_list_old=util.readLinesFromFile(old_train_txt);
tif_list_old=[line[line.index(' ')+1:] for line in tif_list_old];
return tif_list_old;
def getDirMetaFromTifPath(tif_path):
dir_meta_old=tif_path.split('/');
dir_meta_old=dir_meta_old[:-3];
dir_meta_old='/'.join(dir_meta_old);
return dir_meta_old;
def getImageDirFromTifPath(tif_path):
image_dir=tif_path.split('/');
image_dir=image_dir[-2];
return image_dir;
def script_sanityCheckEquivalenceClusters():
old_train_txt='/disk2/mayExperiments/ft_youtube_hmdb_newClusters/train.txt';
new_train_txt='/disk3/maheen_data/ft_youtube_40/train.txt';
tif_list_old=getTifsFromTrain(old_train_txt);
tif_list_new=getTifsFromTrain(new_train_txt);
print len(tif_list_old),len(tif_list_new);
dir_meta_old=getDirMetaFromTifPath(tif_list_old[0]);
dir_meta_new=getDirMetaFromTifPath(tif_list_new[0]);
image_dir_old=getImageDirFromTifPath(tif_list_old[0]);
image_dir_new=getImageDirFromTifPath(tif_list_new[0]);
print dir_meta_old,dir_meta_new,image_dir_old,image_dir_new
# return
im_names_old=util.getFileNames(tif_list_old);
im_names_new=util.getFileNames(tif_list_new);
tif_list_both=list(set(im_names_old).intersection(set(im_names_new)));
print len(tif_list_both)
num_to_pick=100;
random.shuffle(tif_list_both);
tif_list_both=tif_list_both[:num_to_pick];
for tif_name in tif_list_both:
video_name=tif_name[:tif_name.index('.')];
old_tif_path=os.path.join(dir_meta_old,video_name,image_dir_old,tif_name);
new_tif_path=os.path.join(dir_meta_new,video_name,image_dir_new,tif_name);
tif_new=scipy.misc.imread(new_tif_path);
tif_old=scipy.misc.imread(old_tif_path);
assert np.array_equal(tif_new,tif_old);
def script_visualizeRatios():
ratio_file='/disk2/mayExperiments/ft_youtube_hmdb_newClusters_layerMagic/ratios.p';
out_file_plot='/disk2/mayExperiments/ft_youtube_hmdb_newClusters_layerMagic/ratios_plot.png';
ratio=pickle.load(open(ratio_file,'rb'));
print ratio.keys();
xAndYs=[];
legend_entries=[];
for key_curr in ratio.keys():
print key_curr,np.array(ratio[key_curr]).shape;
list_curr=np.array(ratio[key_curr]);
index_nan=np.min(np.where(np.isnan(list_curr)));
assert np.sum(np.isnan(list_curr[index_nan:]))==list_curr[index_nan:].size
list_curr=list_curr[:index_nan];
list_curr=list_curr[:100]
xAndYs.append((range(len(list_curr)),list_curr));
legend_entries.append(key_curr);
visualize.plotSimple(xAndYs,out_file_plot,'update/weight ratio','iterations','ratio',legend_entries,0,True);
def script_compareClusters():
clusters_me='/disk2/mayExperiments/youtube_subset_new_cluster/clusters.mat'
clusters_j='/home/maheenrashid/Downloads/debugging_jacob/optical_flow_prediction_test/examples/opticalflow/clusters.mat';
out_file='/disk2/temp/clusters_comp.png';
# clusters_me=scipy.io.loadMat(cluters_me);
clusters_me=scipy.io.loadmat(clusters_me)['C'];
mags_me=np.linalg.norm(clusters_me,axis=1);
print mags_me.shape;
print np.mean(mags_me);
# clusters_me=clusters_me*4;
clusters_j=po.readClustersFile(clusters_j);
mags_j=np.linalg.norm(clusters_j,axis=1);
print mags_j.shape;
print np.mean(mags_j);
return
print clusters_me
print clusters_j
xAndYs=[(clusters_me[:,0],clusters_me[:,1]),(clusters_j[:,0],clusters_j[:,1])]
visualize.plotScatter(xAndYs,out_file,color=['r','b']);
def script_seeMultipleClusters(dir_clusters=None,out_dir_plot=None):
if dir_clusters is None:
dir_clusters='/disk3/maheen_data/debug_networks/clusters_youtube_multiple';
if out_dir_plot is None:
out_dir_plot='/disk2/temp/cluster_plots';
util.mkdir(out_dir_plot);
clusters_all=util.getFilesInFolder(dir_clusters,'.npy');
print len(clusters_all);
for idx_cluster_file,cluster_file in enumerate(clusters_all):
print idx_cluster_file;
cluster_name=util.getFileNames([cluster_file],ext=False)[0];
out_file=os.path.join(out_dir_plot,cluster_name+'.png');
cluster_curr=np.load(cluster_file);
visualize.plotScatter([(cluster_curr[:,0],cluster_curr[:,1])],out_file,color='r');
# files_all.append(out_file);
visualize.writeHTMLForFolder(out_dir_plot,ext='.png',height=300,width=300);
def scaleAndSingleTif(tif):
tif=tif[:,:,0];
tif=tif*255;
tif=np.dstack((tif,tif,tif));
return tif;
def script_checkSuppressFlowMatlabCode():
dir_meta='/disk2/temp/aeroplane_10_3';
dir_bef=os.path.join(dir_meta,'noThresh');
dir_aft=os.path.join(dir_meta,'withThresh');
out_dir=os.path.join(dir_meta,'viz');
util.mkdir(out_dir)
tif_files=util.getFilesInFolder(dir_bef,'.tif');
tif_files=util.getFileNames(tif_files);
for file_curr in tif_files:
file_bef=os.path.join(dir_bef,file_curr);
file_aft=os.path.join(dir_aft,file_curr);
tif_bef_one=scipy.misc.imread(file_bef);
tif_bef_one=tif_bef_one[:,:,0];
tif_aft_one=scipy.misc.imread(file_aft);
tif_aft_one=tif_aft_one[:,:,0];
mat_info_bef=scipy.io.loadmat(os.path.join(dir_bef,file_curr[:file_curr.rindex('.')]+'.mat'));
R=mat_info_bef['R'];
L=mat_info_bef['L'];
# print optFlow.shape,R.shape,L.shape
# optFlow=np.dstack((optFlow,np.zeros((optFlow.shape[0],optFlow.shape[1],1))));
# optFlow=cv2.resize(optFlow,(20,20));
# optFlow=cv2.resize(optFlow,(R.shape[1],R.shape[0]));
# print optFlow.shape
# mag_bef_o=np.power(np.power(optFlow[:,:,0],2)+np.power(optFlow[:,:,1],2),0.5);
mag_bef=np.power(np.power(R,2)+np.power(L,2),0.5);
idx=np.where(mag_bef<1.0);
# print idx[0].shape;
# idx_o=np.where(mag_bef_o<1.0);
# print idx_o[0].shape
# print np.setdiff1d(idx[0],idx_o[0])
# print np.setdiff1d(idx[1],idx_o[1])
# break;
print 'BEFORE'
print np.unique(R[idx]);
print np.unique(tif_bef_one[idx]);
print 'AFTER'
mat_info_aft=scipy.io.loadmat(os.path.join(dir_aft,file_curr[:file_curr.rindex('.')]+'.mat'));
print np.unique(mat_info_aft['R'][idx]);
print np.unique(tif_aft_one[idx]);
assert np.unique(tif_aft_one[idx])[0]==40;
def script_makeUCFTestTrainTxt():
dir_meta='/home/maheenrashid/Downloads/opticalflow/videos/v_BabyCrawling_g01_c01/images';
out_dir='/disk3/maheen_data/debug_networks/sanityCheckDebug';
util.mkdir(out_dir);
train_file=os.path.join(out_dir,'train.txt');
tifs=util.getFilesInFolder(dir_meta,'.tif');
imgs=[file_curr.replace('.tif','.jpg') for file_curr in tifs];
for file_curr in imgs:
assert os.path.exists(file_curr)
lines=[img+' '+tif for img,tif in zip(imgs,tifs)];
util.writeFile(train_file,lines);
def main():
dir_clusters='/disk2/temp/youtube_clusters_check_nothresh';
out_dir=os.path.join(dir_clusters,'viz');
util.mkdir(out_dir);
script_seeMultipleClusters(dir_clusters,out_dir)
return
script_seeMultipleClusters();
return
dir_clusters='/disk3/maheen_data/debug_networks/clusters_youtube_multiple';
clusters_all=util.getFilesInFolder(dir_clusters,'.npy');
clusters_all=[file_curr for file_curr in clusters_all if 'harder' in file_curr];
clusters_all.append(os.path.join(dir_clusters,'clusters_original.npy'));
min_mags=[];
for file_curr in clusters_all:
clusters=np.load(file_curr);
mags=np.power(np.sum(np.power(clusters,2),axis=1),0.5);
min_mag=np.min(mags);
min_mags.append(min_mag);
print min_mags,np.max(min_mags);
thresh=1;
counts=[];
for file_curr in clusters_all:
clusters=np.load(file_curr);
print file_curr
mags=np.power(np.sum(np.power(clusters,2),axis=1),0.5);
count=np.sum(mags<=thresh);
print count
counts.append(count);
print np.mean(counts);
# return
dir_curr='/disk3/maheen_data/debug_networks/figuringClustering';
mag_file=os.path.join(dir_curr,'mags_all.npy');
mags=np.load(mag_file);
print len(mags),np.sum(mags<=thresh);
mags=mags[mags>thresh];
print len(mags);
out_file=os.path.join(dir_curr,'mag_hist_noZero.png');
visualize.hist(mags,out_file,bins=40,normed=True,xlabel='Value',ylabel='Frequency',title='',cumulative=False);
print out_file.replace('/disk3','vision3.cs.ucdavis.edu:1001');
print np.min(mags),np.max(mags),np.mean(mags),np.std(mags);
# def plotSimple(xAndYs,out_file,title='',xlabel='',ylabel='',legend_entries=None,loc=0,outside=False):
if __name__=='__main__':
main();