def createFeatures(subjects=None, resample_dim=[4, 4, 4]):
clearDB()
if os.path.isfile('/code/neurovault/apps/statmaps/tests/features'+str(subjects)+str(resample_dim)+'.npy'): #and subjects == None:
return np.load('/code/neurovault/apps/statmaps/tests/features'+str(subjects)+str(resample_dim)+'.npy').T, \
pickle.load(open('/code/neurovault/apps/statmaps/tests/dict_feat'+str(subjects)+str(resample_dim)+'.p',"rb" ))
else:
feature_dimension = get_feature_dimension(resample_dim)
features = np.empty([feature_dimension, subjects]) #4*4*4 = 28549 #16*16*16 = 450
dict_feat = {}
u1 = User.objects.create(username='******')
for i, file in enumerate(os.listdir('/code/neurovault/apps/statmaps/tests/bench/images/')):
# print 'Adding subject ' + file
print i
randomCollection = Collection(name='random' + file, owner=u1, DOI='10.3389/fninf.2015.00008' + str(i))
randomCollection.save()
image = save_statmap_form(image_path=os.path.join('/code/neurovault/apps/statmaps/tests/bench/images/', file),
collection=randomCollection, image_name=file, ignore_file_warning=True)
if not image.reduced_representation or not os.path.exists(image.reduced_representation.path):
image = save_resampled_transformation_single(image.pk, resample_dim)
features[:, i] = np.load(image.reduced_representation.file)
dict_feat[i] = int(file.split(".")[0])
if i == subjects-1:
features[np.isnan(features)] = 0
np.save('/code/neurovault/apps/statmaps/tests/features'+str(subjects)+str(resample_dim)+'.npy', features)
pickle.dump(dict_feat,open('/code/neurovault/apps/statmaps/tests/dict_feat'+str(subjects)+str(resample_dim)+'.p',"wb" ))
return features.T, dict_feat
def handle(self, *args, **options):
down_data()
clearDB()
app_path = '/code/neurovault/apps/statmaps/tests/bench'
u1 = User.objects.create(username='******')
num_files = len(os.listdir(os.path.join(app_path, 'images/')))
index_table = np.zeros(num_files)
query_table = np.zeros(num_files)
for i, file in enumerate(os.listdir(os.path.join(app_path, 'images/'))):
#print 'Adding subject ' + file
randomCollection = Collection(name='random' + file, owner=u1, DOI='10.3389/fninf.2015.00008' + str(i))
randomCollection.save()
t = Timer()
with t:
image = save_statmap_form(
image_path=os.path.join(app_path, 'images/', file),
collection=randomCollection,
image_name=file,
ignore_file_warning=True)
index_table[i] = t.interval
np.save(os.path.join(app_path, 'results_index'+ str(datetime.datetime.utcnow())[:10]), index_table)
factory = RequestFactory()
request = factory.get('/images/' + str(i) + '/find_similar')
t = Timer()
with t:
_dummy = find_similar(request, image.pk)
query_table[i] = t.interval
np.save(os.path.join(app_path, 'results_query' + str(datetime.datetime.utcnow())[:10]), query_table)
def handle(self, *args, **options):
down_data()
clearDB()
User.objects.all().delete()
app_path = '/code/neurovault/apps/statmaps/tests/bench'
u1 = User.objects.create(username='******')
num_files = len(os.listdir(os.path.join(app_path, 'images/')))
index_table = np.zeros(num_files)
query_table = np.zeros(num_files)
for i, file in enumerate(os.listdir(os.path.join(app_path, 'images/'))):
#print 'Adding subject ' + file
randomCollection = Collection(name='random' + file, owner=u1, DOI='10.3389/fninf.2015.00008' + str(i))
randomCollection.save()
t = Timer()
with t:
image = save_statmap_form(
image_path=os.path.join(app_path, 'images/', file),
collection=randomCollection,
image_name=file,
ignore_file_warning=True)
index_table[i] = t.interval
np.savetxt(os.path.join(app_path, 'results_index'+ str(datetime.datetime.utcnow())[:10] + '.csv'),
index_table, delimiter=",")
t = Timer()
with t:
_dummy = get_similar_images(int(image.pk))
query_table[i] = t.interval
np.savetxt(os.path.join(app_path, 'results_query' + str(datetime.datetime.utcnow())[:10] + '.csv'),
query_table, delimiter=",")
def get_feature_dimension(resample_dim):
u1 = User.objects.create(username='******'+str(resample_dim))
for file in os.listdir('/code/neurovault/apps/statmaps/tests/bench/images/'):
randomCollection = Collection(name='random' + file, owner=u1, DOI='10.3389/fninf.2015.00008' + file)
randomCollection.save()
image = save_statmap_form(image_path=os.path.join('/code/neurovault/apps/statmaps/tests/bench/images/', file),
collection=randomCollection, image_name=file, ignore_file_warning=True)
if not image.reduced_representation or not os.path.exists(image.reduced_representation.path):
image = save_resampled_transformation_single(image.pk, resample_dim)
feature = np.load(image.reduced_representation.file)
dimension = feature.shape[0]
clearDB()
break
return dimension
def handle(self, *args, **options):
clearDB()
app_path = '/code/neurovault/apps/statmaps/tests/bench'
u1 = User.objects.create(username='******')
dict_feat = {}
for i, file in enumerate(os.listdir(os.path.join(app_path, 'images/'))):
#print 'Adding subject ' + file
randomCollection = Collection(name='random' + file, owner=u1, DOI='10.3389/fninf.2015.00008' + str(i))
randomCollection.save()
t = Timer()
with t:
image = save_statmap_form(
image_path=os.path.join(app_path, 'images/', file),
collection=randomCollection,
image_name=file,
ignore_file_warning=True)
dict_feat[i] = image.pk
print i
if i == 939:
pickle.dump(dict_feat, open('/code/neurovault/apps/statmaps/tests/dict_feat_localhost.p', "wb"))
break
def tearDown(self):
clearDB()
def tearDown(self):
clearDB()
def tearDown(self):
shutil.rmtree(self.tmpdir)
clearDB()
def tearDown(self):
shutil.rmtree(self.tmpdir)
clearDB()
def tearDownClass(cls):
clearDB()
cls.user.delete()
def tearDownClass(cls):
clearDB()
cls.user.delete()