def load_model():
sess = tf.InteractiveSession()
_, predictModel = dbnet()
model = predictModel
model.load_weights('models/model.h5', by_name=True, skip_mismatch=True)
config = CropConfig("card", 1, 1)
model_crop = modellib.MaskRCNN(mode="inference",
model_dir="logs",
config=config)
model_crop.load_weights("model_card/mask_rcnn_card_0030.h5", by_name=True)
graph = tf.get_default_graph()
return model, graph, sess, model_crop
if sside < 5:
continue
box[:, 0] = np.clip(np.round(box[:, 0] / width * dest_width), 0,
dest_width)
box[:, 1] = np.clip(np.round(box[:, 1] / height * dest_height), 0,
dest_height)
boxes.append(box.tolist())
scores.append(score)
return boxes, scores
if __name__ == '__main__':
mean = np.array([103.939, 116.779, 123.68])
_, model = dbnet()
model.load_weights(
'/home/nduforet/Projects/Notebooks/DifferentiableBinarization/checkpoints/2020-01-29/simpler_01_0.4214_0.7732.h5',
by_name=True,
skip_mismatch=True)
for image_path in glob.glob(
osp.join('datasets/total_text_subsample/test_images', '*.jpg')):
image = cv2.imread(image_path)
src_image = image.copy()
h, w = image.shape[:2]
image = resize_image(image)
image = image.astype(np.float32)
image -= mean
image_input = np.expand_dims(image, axis=0)
p = model.predict(image_input)[0]
print(p)
if not osp.exists(checkpoints_dir):
os.makedirs(checkpoints_dir)
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
session = tf.Session(config=config)
train_generator = generate(
'/new_home/xxiao/DB_with_normal_DCN/datasets/report',
batch_size=batch_size,
is_training=True)
val_generator = generate('/new_home/xxiao/DB_with_normal_DCN/datasets/report',
batch_size=batch_size,
is_training=False)
model, prediction_model = dbnet()
# resnet_filename = 'ResNet-50-model.keras.h5'
# resnet_resource = 'https://github.com/fizyr/keras-models/releases/download/v0.0.1/{}'.format(resnet_filename)
# resnet_filepath = get_file('/new_home/xxiao/DifferentiableBinarization_TF/models/db_48_2.0216_2.5701.h5', resnet_resource, cache_subdir='models',
# md5_hash='3e9f4e4f77bbe2c9bec13b53ee1c2319')
model.load_weights(
'/new_home/xxiao/DifferentiableBinarization_TF/models/first_1.7646_2.0746.h5',
by_name=True,
skip_mismatch=True)
# model.compile(optimizer=optimizers.Adam(lr=1e-3), loss={'db_loss': lambda y_true, y_pred: y_pred})
model.compile(optimizer=optimizers.SGD(lr=1e-4, momentum=0.8, decay=1e-6),
loss={
'db_loss': lambda y_true, y_pred: y_pred
})
checkpoint = callbacks.ModelCheckpoint(osp.join(
checkpoints_dir, 'db_{epoch:02d}_{loss:.4f}_{val_loss:.4f}.h5'),