def test_decode_raw_data(image_src):
#setup
file_handler = open(image_src, 'rb') # access file data
raw_data = file_handler.read() # read the binary content
config_filepath = model_constants['config']
weights_filepath = model_constants['weights']
status = None
try:
image_descriptor = Descriptor(config_filepath, weights_filepath)
except Exception as e:
status = e
#compute
decoded_image = image_descriptor.decode_raw_data(raw_data)
#verification
assert status is None
assert len(decoded_image.shape) == 3
def test_compute(image_src):
#setup
file_handler = open(image_src, 'rb') # access file data
raw_data = file_handler.read() # read the binary content
config_filepath = model_constants['config']
weights_filepath = model_constants['weights']
status = None
try:
image_descriptor = Descriptor(config_filepath, weights_filepath)
except Exception as e:
status = e
#compute
stacked_output, scale_factor = image_descriptor.compute(raw_data)
#verification
assert status is None
assert stacked_output[:, 5:].shape[1] == 80 # 80 coco classes dataset
assert np.all(scale_factor == np.array([640, 480, 640, 480]))
def test_if_out_of_image_shape(width, height, nb_coord):
#setup
x = np.random.randint(-width, width, size=(nb_coord, 1))
y = np.random.randint(-height, height, size=(nb_coord, 1))
w = np.random.randint(width // 2, width, size=(nb_coord, 1))
h = np.random.randint(height // 2, height, size=(nb_coord, 1))
matrix = np.block([x, y, w,
h]) # horizontal concatenation : shape (nb_cord, 4)
config_filepath = model_constants['config']
weights_filepath = model_constants['weights']
status = None
try:
image_descriptor = Descriptor(config_filepath, weights_filepath)
except Exception as e:
status = e
#compute
coordinates = image_descriptor.check_if_out_of_image_shape(
matrix, width, height)
#verification
assert coordinates.shape == (nb_coord, 4) # shape conservation
assert np.all(coordinates[:, 2:] <= np.array([width, height])
) == True # rescale wh for each object
assert np.all(coordinates[:, :2] >= 0) == True # remove negative position
def test_create_descriptor_success(thr, nms):
#setup
config_filepath = model_constants['config']
weights_filepath = model_constants['weights']
#compute
status = None
try:
image_descriptor = Descriptor(config_filepath, weights_filepath)
except Exception as e:
status = e
#verification
assert status is None
assert isinstance(image_descriptor, Descriptor)
assert len(image_descriptor.layers) == model_constants['nb_layers']
assert len(image_descriptor.out_layers) == 3
def test_create_descriptor_faillure(config_filename, weights_filename):
#setup
current_dir = path.dirname(path.realpath(__file__))
config_filepath = path.join(current_dir, '..', f'models/{config_filename}')
weights_filepath = path.join(current_dir, '..',
f'models/{weights_filename}')
#compute
status = None
try:
image_descriptor = Descriptor(config_filepath, weights_filepath)
except Exception as e:
status = e
#verification
assert status is not None
assert isinstance(status, FileNotFoundError)
def test_object_detection(image_src):
#setup
file_handler = open(image_src, 'rb') # access file data
raw_data = file_handler.read() # read the binary content
config_filepath = model_constants['config']
weights_filepath = model_constants['weights']
status = None
try:
image_descriptor = Descriptor(config_filepath, weights_filepath)
except Exception as e:
status = e
#compute
result = image_descriptor.detect(raw_data)
result_keys = list(result.keys())
result_values = list(result.values())
result_labels = list(
map(lambda idx: model_constants['labels.names'][idx],
result['labels']))
#verification
assert status is None
assert result[
'status'] == 0 # the detection process was performed successfully
assert result_keys == ['status', 'coords', 'scores', 'labels']
assert len(set(result_labels) & set(['car', 'dog', 'aeroplane'])) > 0
from os import path
from flask import Flask
from logger.log import logger
from descriptor.descript import Descriptor
current_dir = path.dirname(path.realpath(__file__))
config_filename = path.join(current_dir, '..', 'models/yolov3.cfg')
weights_filename = path.join(current_dir, '..', 'models/yolov3.weights')
image_descriptor = None
try:
image_descriptor = Descriptor(config_filename, weights_filename, 0.5, 0.3)
logger.info('the descriptor is ready')
except FileNotFoundError as contents:
logger.error(
'some files are missing, please check the weight or config file')
except Exception as e:
logger.error(e)
max_row = 1024
max_col = 1024
max_channels = 3
extentions = ['jpg', 'gif', 'png', 'ttf', 'jpeg']
app = Flask(__name__)
app.config['MAX_CONTENT_LENGTH'] = max_row * max_col * max_channels
app.config['VALID_EXTENSIONS'] = extentions
app.config['IMAGE_DESCRIPTOR'] = image_descriptor