def getTrainLowFeature():
rootDir = "E:/ImageDataset/train/train_low"
paths, counts = getPath.getPath(rootDir)
snap = np.zeros((100, 100))
i = 5263
for path in paths:
print 'trainlow ' + path
image = cv2.imread(path)
resize = extractEdge(image)
np.save('E:/color/data/spatialDistributionofEdges/{0}.npy'.format(i),
resize)
i = i + 1
snap = snap + resize
counts = float(counts)
print counts
Ms = snap / counts
return Ms
def getTrainHighFeature():
print 'hello'
rootDir = "E:/ImageDataset/train/train_high"
paths, counts = getPath.getPath(rootDir)
pro = np.zeros((100, 100))
i = 1
for path in paths:
print 'trainhigh ' + path
image = cv2.imread(path)
resize = extractEdge(image)
np.save('E:/color/data/spatialDistributionofEdges/{0}.npy'.format(i),
resize)
i = i + 1
pro = pro + resize
counts = float(counts)
Mp = pro / counts
return Mp
from sklearn.grid_search import GridSearchCV
import getPath
from sklearn.cross_validation import cross_val_score
ntrain_high=2262
ntrain_low=6581
ntest_high=2262
ntest_low=6580
ntrain = 8843
ntest = 8842
count = 17685
feature_dim=7
feature_train=np.zeros((feature_dim,ntrain))
root_train='E:/featureData_CUHK/train'
paths_train,count_train=getPath.getPath(root_train)
i=0
for path in paths_train:
feature=np.load(path)
feature=np.array(feature)
feature_train[i]=feature
i=i+1
train_feature=np.transpose(feature_train) #求出训练集的特征
label_train=np.array([])
for i in range(1,ntrain_high+1):
label_train=np.append(label_train,1)
for j in range(1,ntrain_low+1):
label_train=np.append(label_train,0)
train_label=np.transpose(label_train)
"""
import getPath
#import colorPalette
import layoutComposition
import edgeComposition
import GT_layout
import GT_edge
import blur
import dark
import Contrasts
import HSVcounts
root_train = 'E:/ImageDataset_AVA/train/'
root_test = 'E:/ImageDataset_AVA/test/'
paths_train, counts_train = getPath.getPath(root_train)
paths_test, counts_test = getPath.getPath(root_test)
root_trainhigh = 'E:/ImageDataset_AVA/train/train_high'
root_trainlow = 'E:/ImageDataset_AVA/train/train_low'
root_testhigh = 'E:/ImageDataset_AVA/test/test_high'
root_testlow = 'E:/ImageDataset_AVA/test/test_low'
paths_trainhigh, counts_trainhigh = getPath.getPath(root_trainhigh)
paths_trainlow, counts_trainlow = getPath.getPath(root_trainlow)
paths_testhigh, counts_testhigh = getPath.getPath(root_testhigh)
paths_testlow, counts_testlow = getPath.getPath(root_testlow)
layoutComposition.layout(paths_trainhigh, paths_testhigh, paths_trainlow,
paths_testlow, paths_train, paths_test)
edgeComposition.EC(paths_trainhigh, paths_testhigh, paths_trainlow,
import matplotlib.pyplot as plt
from sklearn.metrics import roc_curve, auc, accuracy_score
import getPath
ntrain_high = 12771
ntrain_low = 12771
ntest_high = 12771
ntest_low = 12771
ntrain = 25542
ntest = 25542
count = 51084
feature_dim = 24
feature_train = np.zeros((feature_dim, ntrain))
root_train = 'E:/efficiency_AVA/data/train'
paths_train, count_train = getPath.getPath(root_train)
i = 0
for path in paths_train:
feature = np.load(path)
feature = np.array(feature)
feature_train[i] = feature
i = i + 1
train_feature = np.transpose(feature_train)
np.save('E:/efficiency_AVA/data/trainfeature.npy', train_feature)
feature_test = np.zeros((feature_dim, ntest))
root_test = 'E:/efficiency_AVA/data/test'
paths_test, count_test = getPath.getPath(root_test)
i = 0
for path in paths_test:
feature = np.load(path)
init_job = ShortestPathIter(args=[graph, '--source', source, '--destination', destination, '--weighted', weighted,
'-r', mode, '--output-dir', 'hdfs:///user/leiyang/out'])
iter_job = ShortestPathIter(args=['hdfs:///user/leiyang/in/part*', '--source', source, '--destination', destination,
'--weighted', weighted, '-r', mode, '--output-dir', 'hdfs:///user/leiyang/out'])
else:
init_job = LongestPathIter(args=[graph, '--source', source, '--weighted', weighted,
'-r', mode, '--output-dir', 'hdfs:///user/leiyang/out'])
iter_job = LongestPathIter(args=['hdfs:///user/leiyang/in/part*', '--source', source,
'--weighted', weighted, '-r', mode, '--output-dir', 'hdfs:///user/leiyang/out'])
if isLongest:
path_job = getLongestDistance(args=['hdfs:///user/leiyang/out/part*', '-r', mode])
else:
path_job = getPath(args=['hdfs:///user/leiyang/out/part*', '--destination', destination, '-r', mode])
if isWeighted or isLongest:
stop_job = isTraverseCompleted(args=['hdfs:///user/leiyang/out/part*', 'hdfs:///user/leiyang/in/part*', '-r', mode])
else:
stop_job = isDestinationReached(args=['hdfs:///user/leiyang/out/part*', '--destination', destination, '-r', mode])
# run initialization job
with init_job.make_runner() as runner:
print str(datetime.datetime.now()) + ': starting initialization job ...'
runner.run()
# move the result to input folder
print str(datetime.datetime.now()) + ': moving results for next iteration ...'
call(['hdfs', 'dfs', '-mv', '/user/leiyang/out', '/user/leiyang/in'])
# run BFS iteratively
# cattles = {'811':'811'}
#VA = {'817':'57.08'}
day = 19
#day = [6, 5, 18, 2]
# day = [22]
low = 9
high = 48
VedioFlag = True #is used for check if there is 'avi' file inside the folder or just the pictures
for llx in cattles:
#pathNameGroup = getPath('/Volumes/WA03-1/WA03 2016 4/' + str(llx), 11, [10, 18, 21], [16, 43, 34])
#pathNameGroup = getPath.getPath('/Volumes/WA03-1/WA03 2016 5/' + str(llx), 9, [10, 18, 21], [16, 43, 34])
#pathNameGroup = getPath.getPath('/Volumes/WA02-4/finalset WA02 2016 1 34013s/' + llx, 19, [0, 0, 0], [0, 0, 0])
pathNameGroup = getPath.getPath(input_path + llx, day, low, high)
#pathNameGroup = getPath('/Volumes/WA03-1/WA03 2016 1/833', 19, [10, 18, 21], [16, 43, 34])
#print(pathNameGroup)
for count in range(len(pathNameGroup)):
#for count in range(0,1):
pathNameBack = pathNameGroup[count]
#pathName = '/Volumes/WA03-1/WA03 2016 1/833/' + pathNameBack
#pathName = '/Volumes/WA03-1/WA03 2016 4/' + str(llx) + '/' + pathNameBack
#pathName = '/Volumes/WA03-1/WA03 2016 5/' + str(llx) + '/' + pathNameBack
#pathName = '/Volumes/WA02-4/finalset WA02 2016 1 34013s/' + llx + '/' + pathNameBack
pathName = input_path + llx + '/' + pathNameBack
#print(pathName)
@author: Administrator
"""
import numpy as np
import getPath
import space
"""import boundingbox
import colorDistribution
import hueCount
import blur
import contrast
import bright"""
root1='E:/ImageDataset_CUHK/test'
root2='E:/ImageDataset_AVA/test'
AUHK_test,counts_test=getPath.getPath(root1)
np.save('E:/path/CUHK_test.npy',AUHK_test)
AVA_test,counts_test=getPath.getPath(root2)
np.save('E:/path/AVA_test.npy',AVA_test)
"""
root='E:/ImageDataset_CUHK/test'
root_train = 'E:/ImageDataset/train/'
root_test = 'E:/ImageDataset/test/'
paths_train,counts_train=getPath.getPath(root_train)
paths_test,counts_test=getPath.getPath(root_test)
print counts_train
print counts_test"""
"""
qedge_train=space.calculateQuality(paths_train) #边缘空间分布特征
qedge_test=space.calculateQuality(paths_test)
def test(request):
start = request.POST['start']
end = request.POST['end']
l = getPath.getPath(start, end)
return render_to_response('index/example2.html', {'nodelist': l},
context_instance=RequestContext(request))
V = V / 255.0
hist = np.zeros((20, 1))
for i in range(height):
for j in range(width):
if S[i][j] > 0.2 and V[i][j] >= 0.15 and V[i][j] <= 0.95:
k = H[i][j] / 18
hist[k][0] = hist[k][0] + 1
return hist
def hueCount(paths):
q_hue = np.array([])
for path in paths:
print 'processing ' + path
image = cv2.imread(path)
hist = calcHist(image)
m = hist.max()
N = sum(hist > alp * m)
qh = 20 - N
q_hue = np.append(q_hue, qh)
return q_hue
root_train = 'E:/ImageDataset/train'
root_test = 'E:/ImageDataset/test'
#paths_train,counts_train=getPath.getPath(root_train)
paths_test, counts_test = getPath.getPath(root_test)
#qhue_train=hueCount(paths_train)
#np.save('E:/featureData/train/qhue_train.npy',qhue_train)
qhue_test = hueCount(paths_test)
np.save('E:/featureData/test/qhue_test.npy', qhue_test)