def update_request(message):
print('Receive', message)
name = message['name']
try:
if name == 'landmarks': update_landmark()
if name == 'model': update_model()
if name == 'lfw': update_lfw()
if name == 'output': update_output()
if name == 'prediction': make_tests()
except Exception as e:
socketio_app.emit('finish-' + str(name),
{'error': "{}: {}".format(type(e).__name__, str(e))})
def check_request(message):
name = message['name']
try:
info = None
if name == 'landmarks': info = get_landmark_info()
if name == 'model': info = get_model_info()
if name == 'lfw': info = get_lfw_info()
if name == 'output': info = get_output_info()
#if name == 'prediction': return
if info:
socketio_app.emit('finish-' + str(name), info)
except Exception as e:
socketio_app.emit('finish-' + str(name),
{'error': "{}: {}".format(type(e).__name__, str(e))})
def download_and_extract_model(model_name, data_dir):
if not os.path.exists(data_dir):
os.makedirs(data_dir)
file_id = model_dict[model_name]
destination = os.path.join(data_dir, model_name + '.zip')
if not os.path.exists(destination):
print('Downloading model to %s' % destination)
socketio_app.emit('log-landmark',
{'message': 'Downloading model to %s' % destination
}) # move emits to logging
download_file_from_google_drive(file_id, destination)
with zipfile.ZipFile(destination, 'r') as zip_ref:
print('Extracting model to %s' % data_dir)
socketio_app.emit('log-landmark',
{'message': 'Extracting model to %s' % data_dir
}) # move emits to logging
zip_ref.extractall(data_dir)
def update_lfw():
socketio_app.emit('log-lfw', {'message': 'Start downloading...'})
archive = absolute(app.config['FACE_NET_DATA_DIR']) + '.tar.gz'
data = absolute(app.config['FACE_NET_DATA_DIR'])
urllib.request.urlretrieve(app.config['FACE_NET_LWF_URL'], archive)
socketio_app.emit('log-lfw', {'message': 'Extracting...'})
os.makedirs(data, exist_ok=True)
tar = tarfile.open(archive, "r:gz")
tar.extractall(data)
tar.close()
os.remove(archive)
socketio_app.emit('log-lfw', {'message': 'Done'})
socketio_app.emit('finish-lfw', get_lfw_info())
def preprocess(input_dir, output_dir, crop_dim):
start_time = time.time()
pool = mp.Pool(processes=mp.cpu_count())
if not os.path.exists(output_dir):
os.makedirs(output_dir)
for image_dir in os.listdir(input_dir):
image_output_dir = os.path.join(output_dir, os.path.basename(os.path.basename(image_dir)))
if not os.path.exists(image_output_dir):
os.makedirs(image_output_dir)
image_paths = glob.glob(os.path.join(input_dir, '**/*.jpg'))
for index, image_path in enumerate(image_paths):
image_output_dir = os.path.join(output_dir, os.path.basename(os.path.dirname(image_path)))
output_path = os.path.join(image_output_dir, os.path.basename(image_path))
pool.apply_async(preprocess_image, (image_path, output_path, crop_dim))
pool.close()
pool.join()
logger.info('Completed in {} seconds'.format(time.time() - start_time))
socketio_app.emit('log-lfw', {'message': 'Completed in {} seconds'.format(time.time() - start_time)}) # move emits to logging
def update_landmark():
socketio_app.emit('log-landmark', {'message': 'Start downloading...'})
file = absolute(app.config['FACE_NET_LANDMARKS_FILE'])
archive = file + '.bz2'
urllib.request.urlretrieve(app.config['FACE_NET_LANDMARKS_URL'], archive)
socketio_app.emit('log-landmark', {'message': 'Extracting...'})
with open(file, 'wb') as new_file, bz2.BZ2File(archive, 'rb') as bz2_file:
for bytes in iter(lambda: bz2_file.read(100 * 1024), b''):
#for bytes in file.read():
new_file.write(bytes)
os.remove(archive)
socketio_app.emit('log-landmark', {'message': 'Done'})
socketio_app.emit('finish-landmark', get_landmark_info())
def update_model():
socketio_app.emit('log-model', {'message': 'Start updating...'})
download_and_extract_model('20170511-185253',
absolute(app.config['FACE_NET_WEIGHTS_DIR']))
socketio_app.emit('log-model', {'message': 'Done'})
socketio_app.emit('finish-model', get_model_info())
def make_tests():
crop_dim = 180
print('make_tests')
socketio_app.emit('log-prediction', {'message': 'Start making tests...'})
im1 = '/home/srivoknovski/Python/flask/acme/Networks/FaceNet/data/lfw/Aaron_Peirsol/Aaron_Peirsol_0002.jpg'
im2 = '/home/srivoknovski/Python/flask/acme/Networks/FaceNet/data/lfw/Aaron_Peirsol/Aaron_Peirsol_0004.jpg'
im3 = '/home/srivoknovski/Python/flask/acme/Networks/FaceNet/data/lfw/Aaron_Tippin/Aaron_Tippin_0001.jpg'
with open(im1, "rb") as image_file:
socketio_app.emit('log-prediction', {'image': (image_file.read())})
with open(im2, "rb") as image_file:
socketio_app.emit('log-prediction', {'image': (image_file.read())})
with open(im3, "rb") as image_file:
socketio_app.emit('log-prediction', {'image': (image_file.read())})
socketio_app.emit(
'log-prediction', {
'message':
'preprocessing test images..., Crop dimension {}'.format(crop_dim)
})
images = []
images.append(face_net_instance.process_image(im1, crop_dim))
images.append(face_net_instance.process_image(im2, crop_dim))
images.append(face_net_instance.process_image(im3, crop_dim))
socketio_app.emit('log-prediction', {'message': 'loading model...'})
model_path = absolute(app.config['FACE_NET_WEIGHTS_FILE'])
embs = get_emmbedings(images=images, model_path=model_path)
socketio_app.emit(
'log-prediction', {
'message':
'Model path {} {}'.format(model_path, os.path.getsize(model_path))
})
diff1 = np.linalg.norm(embs[0] - embs[1])
diff2 = np.linalg.norm(embs[0] - embs[2])
print(im1, im2, np.linalg.norm(embs[0] - embs[1]))
print(im1, im3, np.linalg.norm(embs[0] - embs[2]))
socketio_app.emit('log-prediction', {'message': 'Done'})
socketio_app.emit('finish-prediction', {
'The same persons': str(diff1),
'The different persons': str(diff2)
})
def update_output():
socketio_app.emit('log-output', {'message': 'Start updating...'})
output_dir = absolute(app.config['FACE_NET_OUTPUT_DIR'])
preprocess(absolute(app.config['FACE_NET_DATA_DIR']), output_dir, 180)
socketio_app.emit('log-output', {'message': 'Done'})
socketio_app.emit('finish-output', get_output_info())