def run(self):
print "Starting " + self.name
while not exitFlag:
queueLock.acquire()
if not workQueue.empty():
currQueue = self.q.get()
queueLock.release()
# fn = currQueue['fn']
# urls = currQueue['urls']
# kps = []
# for url in urls:
# img = image.loadImageFromUrl(url, cv2.IMREAD_GRAYSCALE, True, 200)
# name = image.fileName(url)
# kp, des = image.keypointDesCalc(img)
# kps.append(image.pack(kp, des, name))
# time.sleep(0)
# image.saveKps(kps, HASH_PATH + fn + KPC_EXT)
fn = currQueue['fn']
urls = currQueue['urls']
kps = []
for url in urls:
img = image.loadImageFromUrl(url, cv2.IMREAD_GRAYSCALE, True, 200)
name = image.fileName(url)
image.keypointDesCalc(img, HASH_PATH + name)
# image.saveKps(kps, HASH_PATH + fn + KPC_EXT)
else:
queueLock.release()
print "Exiting " + self.name
def process_data(threadName, q):
while not exitFlag:
queueLock.acquire()
if not workQueue.empty():
url = q.get()
queueLock.release()
name2 = image.fileName(url)
img2 = image.loadImageFromUrl(url)
image.compare(name1, name2, img1.copy(), img2)
# print "%s processing %s" % (threadName, data)
else:
queueLock.release()
def run(self):
print("Starting " + self.name)
while not exitFlag:
queueLock.acquire()
if not workQueue.empty():
url = self.q.get()
queueLock.release()
img = image.loadImageFromUrl(url, cv2.IMREAD_GRAYSCALE, False)
name = image.fileName(url)
image.keypointDesCalcDb(self.collection, img, name, 100)
else:
queueLock.release()
print("Exiting " + self.name)
def run(self):
print("Starting " + self.name)
while not exitFlag:
self.lock.acquire()
if not self.queue.empty():
data = self.queue.get()
self.lock.release()
url = data[0]
folder = '%s%s/' % (HASH_PATH, data[1])
img = image.loadImageFromUrl(url, resize=True, maxSize=800)
name = image.fileName(url)
image.keypointDesCalc(img, folder + name, 100, self.wlock)
else:
self.lock.release()
print("Exiting " + self.name)
def loadFiles(filesInQueue=10000, hashPath=HASH_PATH, ext=DES_EXT):
def load_descriptors(bow, hashPath=HASH_PATH, withSubFolders=True, ext=DES_EXT):
des_list = []
kp_list = []
folder = 0
path = "%s%s/" % (hashPath, folder)
while os.path.exists(path):
for imagePath in glob.glob("%s*%s" % (path, '.des')):
des_list.append(imagePath)
kp_list.append('%s%s.kp' % (path, im.fileName(imagePath)))
if len(des_list) == FILES_CHECK:
folder = -2
break
folder += 1
path = "%s%s/" % (hashPath, folder)
if len(des_list) == 0:
print("Hash files count is empty")
return []
print("Files count: %d " % len(des_list))
for n in des_list:
bow.add(im.loadDesFromPath(n))
# bow.write(HASH_PATH + 'hash.bat')
print("Loaded")
images = []
for n in des_list:
images.append('http://comicstore.cf/uploads/diamonds/%s.jpg' % im.fileName(n))
return images, kp_list
flann = cv2.FlannBasedMatcher(index_params, search_params)
bow_train = cv2.BOWKMeansTrainer(50) # toy world, you want more.
bow_extract = cv2.BOWImgDescriptorExtractor(sift, flann)
images, kps = load_descriptors(bow_train)
print 'Getting cluster ...'
voc = bow_train.cluster()
print 'Set vocabulary ...'
bow_extract.setVocabulary(voc)
traindata, trainlabels = [], []
j = 0
for i in images:
img = im.loadImageFromUrl(i, resize=True, maxSize=800)
kp = im.loadKeypointFromPath(kps[j])
traindata.extend(bow_extract.compute(img, kp))
trainlabels.append(j)
j += 1
print 'Training SVM'
svm = cv2.ml.SVM_create()
svm.train(np.array(traindata), cv2.ml.ROW_SAMPLE, np.array(trainlabels))
img = im.loadImageFromPath('%s%s' % (IMG_PATH, 'M2.jpg'), resize=True, maxSize=800)
kp, des = im.keypointDesCalc(img, count=100)
fbow = bow_extract.compute(img, kp)
t_start = cv2.getTickCount()
p = svm.predict(fbow)
t_end = cv2.getTickCount()
print "Time match: %s" % ((t_end - t_start) / cv2.getTickFrequency())
def downloadImages():
nameList = sql.allComics()[:1000]
for i in nameList:
img = im.loadImageFromUrl(i, color=cv2.IMREAD_COLOR, resize=False, maxSize=800)
cv2.imwrite('/home/evgeniy/projects/python/open-cv/images/comics/full/%s.jpg' % im.fileName(i), img)