def test(intercept, featureWeight):
testFeatures = getFeatures(0)
region = testFeatures.region
score = []
for i in range(len(region)):
score.append({})
fOut = open("../txt/result.txt",'w')
for i in range(len(region)):
for query in region[i]:
score[i][query] = {}
for url in region[i][query]:
score[i][query][url] = \
dot_product(region[i][query][url],featureWeight) + intercept
url_sorted_by_value = OrderedDict(sorted(score[i][query].items(), key=lambda x: x[1]))
url_sorted_by_value_r = list(reversed(url_sorted_by_value))
out = str(query) + " " + str(i) + " "
for u in url_sorted_by_value_r:
out = out + str(u) + " "
out = out+ "\n"
fOut.write(out)
def consolidate_features(base_graphs, Gcollab_delta, k):
features = {}
Gcollab = base_graphs[graphutils.Graph.COLLAB]
feature_graphs = graphutils.split_feat_graphs(base_graphs)
for node in Gcollab.Nodes():
nodeID = node.GetId()
for neighborID in graphutils.getKHopN(Gcollab, nodeID, k):
if nodeID > neighborID: # swap
nodeID = neighborID + nodeID
neighborID = nodeID - neighborID
nodeID = nodeID - neighborID
if (nodeID, neighborID) in features:
continue
features[(nodeID, neighborID)] = []
for graph in feature_graphs:
features = getFeatures(Gcollab, Gcollab_delta, graph, features)
return features
def consolidate_features(base_graphs, Gcollab_delta, k):
features = {}
Gcollab = base_graphs[graphutils.Graph.COLLAB]
feature_graphs = graphutils.split_feat_graphs(base_graphs)
for node in Gcollab.Nodes():
nodeID= node.GetId()
for neighborID in graphutils.getKHopN(Gcollab, nodeID, k):
if nodeID > neighborID: # swap
nodeID= neighborID + nodeID
neighborID= nodeID - neighborID
nodeID= nodeID - neighborID
if (nodeID, neighborID) in features:
continue
features[(nodeID, neighborID)]= []
for graph in feature_graphs:
features = getFeatures(Gcollab, Gcollab_delta, graph, features)
return features
import matplotlib.pyplot as plt
if __name__ == '__main__':
# setup video capture
cap = cv2.VideoCapture("input_videos/Easy.mp4")
ret,img1 = cap.read()
ret,img2 = cap.read()
cap.release()
maxCorners = 20
qualityLevel = 0.01
minDistance = 8
bbox_list = []
bbox_pts = []
bbox_list, bbox_pts, new_img = getBoundingBox(img1, bbox_list, bbox_pts)
img1_gray = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY)
startXs, startYs, _ = getFeatures(img1_gray, bbox_list, maxCorners, qualityLevel, minDistance)
newXs = startXs + 5
newYs = startYs + 7
print(startXs, startYs)
print(newXs, newYs)
#nnewXs, nnewYs = estimateAllTranslation(newXs, newYs, img2, img3)
#print(len(newXs[0]))
#print(len(newYs[0]))
plt.figure()
plt.imshow(img2)
#plt.plot(newYs, newXs, 'w+')
plt.axis('off')
plt.show()
import time
import commands
from getFeatures import *
from featureProcess import getAverage
from featureProcess import normalize
from rank import *
print '##################################'
print '# Program start #'
print '##################################'
print '-Stage 1- extract features..'
trainStruct = getFeatures(1)
print '-Stage 2- processing features..'
trainStruct = getAverage(trainStruct)
print '-Stage 3- normalizing features..'
trainStruct = normalize(trainStruct)
print '-Stage 4- generating training set for Weka..'
arffGen(trainStruct.region)
time.sleep(1)
print '-Stage 5- getting trained model parameters..'
beta = commands.getoutput("java WekaTester trainClickThrough.arff").split('\n')
intercept = float(beta[0])
beta = beta[1:len(beta)-1]
for i,item in enumerate(beta):
beta[i] = float(item)
print beta[i]
print '-Stage 6- get test set..'
test(intercept,beta)
