h = image.shape[0]
w = image.shape[1]
input_image_shape = tf.constant([h, w], dtype=tf.dtypes.float32)
img_tensor = resize_image_with_pad(image)
img_tensor = tf.dtypes.cast(img_tensor, dtype=tf.dtypes.float32)
# img_tensor = img_tensor / 255.0
yolo_output = model(img_tensor, training=False)
boxes, scores, classes = Inference(
yolo_output=yolo_output,
input_image_shape=input_image_shape).get_final_boxes()
image_with_boxes = draw_boxes_on_image(cv2.imread(image_dir), boxes,
scores, classes)
return image_with_boxes
if __name__ == '__main__':
# GPU settings
gpus = tf.config.list_physical_devices(device_type="GPU")
if gpus:
for gpu in gpus:
tf.config.experimental.set_memory_growth(device=gpu, enable=True)
# load model
yolo_v3 = YOLOV3(out_channels=3 * (CATEGORY_NUM + 5))
yolo_v3.load_weights(filepath=save_model_dir + "saved_model")
# inference
image = single_image_inference(image_dir=test_picture_dir, model=yolo_v3)
cv2.namedWindow("detect result", flags=cv2.WINDOW_NORMAL)
cv2.imshow("detect result", image)
cv2.waitKey(0)
input_shape=[IMAGE_HEIGHT,
IMAGE_WIDTH]).generate_label()
return true_label
if __name__ == '__main__':
# GPU settings
gpus = tf.config.list_physical_devices(device_type="GPU")
if gpus:
for gpu in gpus:
tf.config.experimental.set_memory_growth(device=gpu, enable=True)
# dataset
train_dataset, train_count = generate_dataset()
net = YOLOV3(out_channels=3 * (CATEGORY_NUM + 5))
print_model_summary(network=net)
if load_weights_before_training:
net.load_weights(filepath=save_model_dir +
"epoch-{}".format(load_weights_from_epoch))
print("Successfully load weights!")
else:
load_weights_from_epoch = -1
# loss and optimizer
yolo_loss = YoloLoss()
lr_schedule = tf.keras.optimizers.schedules.ExponentialDecay(
initial_learning_rate=0.001,
decay_steps=5000,
decay_rate=0.96,