def response(text, name):
data = dict()
preds = predict(text, name)
slot_dict = merge(text, preds)
data['content'] = text
data['slot'] = slot_dict
data_str = json.dumps(data, ensure_ascii=False)
logger.info(data_str)
return data_str
def response():
data = request.get_json()
words, preds = predict(data['content'])
entitys, labels = list(), list()
for word, pred in zip(words, preds):
pred = ind_labels[pred]
if pred != 'O':
entitys.append(word)
labels.append(zh_en[pred])
slot_dict = make_dict(entitys, labels)
data['slot'] = slot_dict
data_str = json.dumps(data, ensure_ascii=False)
logger.info(data_str)
return data_str
def rec():
try:
base64_data = request.get_data().strip()
ret_num = int(predict(base64_data))
print(ret_num)
result = {'status': 'success', 'data': ret_num}
except Exception as e:
result = {
'status': 'error',
'data': traceback.format_exc(),
}
response = make_response(json.dumps(result))
response.headers['Access-Control-Allow-Origin'] = '*'
return response
def test(name, sents):
flat_labels, flat_preds = [0], [0]
for text, pairs in sents.items():
labels = list()
for pair in pairs:
labels.append(label_inds[pair['label']])
bound = len(text) - seq_len if len(text) > seq_len else 0
flat_labels.extend(labels[bound:])
flat_preds.extend(predict(text, name))
precs = precision_score(flat_labels, flat_preds, average=None)
recs = recall_score(flat_labels, flat_preds, average=None)
with open(map_item(name, paths), 'w') as f:
f.write('label,prec,rec' + '\n')
for i in range(1, class_num):
f.write('%s,%.2f,%.2f\n' % (ind_labels[i], precs[i], recs[i]))
f1 = f1_score(flat_labels,
flat_preds,
average='weighted',
labels=label_set)
print('\n%s f1: %.2f - acc: %.2f' %
(name, f1, accuracy_score(flat_labels, flat_preds)))
def test(sents):
flat_labels, flat_preds = list(), list()
for text, triples in sents.items():
word1s, labels = list(), list()
for triple in triples:
word1s.append(triple['word'])
labels.append(label_inds[triple['label']])
word2s, preds = predict(text)
for i in range(len(word2s)):
if word2s[i] == word2s[i]:
flat_labels.append(labels[i])
flat_preds.append(preds[i])
precs = precision_score(flat_labels, flat_preds, average=None)
recs = recall_score(flat_labels, flat_preds, average=None)
with open(path_crf, 'w') as f:
f.write('label,prec,rec' + '\n')
for i in range(class_num):
f.write('%s,%.2f,%.2f\n' % (ind_labels[i], precs[i], recs[i]))
f1 = f1_score(flat_labels,
flat_preds,
average='weighted',
labels=label_set)
print('\n%s f1: %.2f - acc: %.2f' %
('crf', f1, accuracy_score(flat_labels, flat_preds)))
time.sleep(0.1)
lbp_face_cascade = cv2.CascadeClassifier(
'opencv-source/data/lbpcascades/lbpcascade_frontalface.xml')
face_recognizer = cv2.face.LBPHFaceRecognizer_create()
face_recognizer.read('data/model.yml')
for frame in camera.capture_continuous(rawCapture, format="bgr"):
image = frame.array
image = np.array(image)
if image is None:
continue
image = rec.predict(image)
cv2.imshow("Frame", image)
key = cv2.waitKey(1) & 0xFF
rawCapture.truncate(0)
if key == ord("q"):
break
#test_img1 = cv2.imread("untrained-data/image1.jpg")
#test_img2 = cv2.imread("untrained-data/image2.jpg")
#test_img3 = cv2.imread("untrained-data/image3.jpg")
#test_img4 = cv2.imread("untrained-data/image4.jpg")
#test_img5 = cv2.imread("untrained-data/image5.jpg")
#test_img6 = cv2.imread("untrained-data/image6.jpg")