class imageTag(object):
fio_r = fio_w = None
api = None
images = []
def __init__(self, input_filename, output_filename):
self.fio_r = open(input_filename, 'r')
self.fio_w = open(output_filename, 'wb')
self.api = ClarifaiApi(model='general-v1.3')
def read_images(self):
""" read in image samples """
worker = Worker(50)
for line in self.fio_r:
line = line.strip()
try:
item = json.loads(line)
except:
continue
worker.add_task(self.tag_image, item)
logging.debug('%d images have been read' % len(items))
worker.join()
def tag_image(self, image):
""" tag and embed one image """
imageurl = image.get('url')
if not imageurl:
return
try:
result = self.api.tag_and_embed_urls(imageurl)
image['tags'] = result['results'][0]['result']['tag']['classes']
image['probs'] = result['results'][0]['result']['tag']['probs']
image['embeds'] = result['results'][0]['result']['embed']
except:
logging.error('FAILED %s' % imageurl)
print 'FINSHED %s' % imageurl
self.fio_w.write(json.dumps(image) + '\n')
def run(self):
""" run the tag service """
self.read_images()