img_size=img_size,
n_imgs=n_imgs)
handle, t_iter, v_iter, images, masks, depths, angles, classes = dataprovider.dataset(
)
# namingyou
model = AE_only_conv()
is_training = tf.placeholder(tf.bool)
# data
normalized_imgs = normalize_images(images)
normalized_masks = normalize_images(masks)
normalized_depths = normalize_images(depths)
cosinized_angles = deg2rad(angles)
relative_angles = cosinized_angles[:, -1, :] # - cosinized_angles[:, 0, :]
base_imgs, target_imgs = split_imgs(normalized_imgs)
base_masks, target_masks = split_imgs(normalized_masks)
_, target_depths = split_imgs(normalized_depths)
base_masks = 1 - base_masks
target_masks = 1 - target_masks
classes = classes[:, 0]
base_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 3])
base_mask_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 1])
target_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 3])
target_depth_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 1])
handle, t_iter, v_iter, images, masks, depths, angles, classes = dataprovider.dataset(
)
# namingyou
model = AE_only_conv()
model_name = '{}---date_{}'.format(note, datetime.datetime.now())
# placeholders
is_training = tf.placeholder(tf.bool)
# data
normalized_imgs = normalize_images(images)
normalized_masks = normalize_images(masks)
normalized_depths = normalize_images(depths)
cosinized_angles = deg2rad(angles)
relative_angles = cosinized_angles[:, -1, :] # - cosinized_angles[:, 0, :]
base_imgs, target_imgs = split_imgs(normalized_imgs)
base_masks, target_masks = split_imgs(normalized_masks)
_, target_depths = split_imgs(normalized_depths)
base_masks = 1 - base_masks
target_masks = 1 - target_masks
classes = classes[:, 0]
base_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 3])
base_mask_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 1])
target_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 3])
target_depth_pl = tf.placeholder(tf.float32, [None, img_size, img_size, 1])