return return_img if not os.path.exists(FLAGS.saved_prediction): os.mkdir(FLAGS.saved_prediction) val_data = input_data.read_val_data(rgb_mean=FLAGS.rgb_mean, crop_height=FLAGS.crop_height, crop_width=FLAGS.crop_width, classes=FLAGS.classes, ignore_label=FLAGS.ignore_label, scales=FLAGS.scales) test_data = input_data.read_test_data(rgb_mean=FLAGS.rgb_mean, crop_height=FLAGS.crop_height, crop_width=FLAGS.crop_width, classes=FLAGS.classes, ignore_label=FLAGS.ignore_label, scales=FLAGS.scales) with tf.name_scope("input"): x = tf.placeholder(tf.float32, [FLAGS.batch_size, FLAGS.height, FLAGS.width, 3], name='x_input') y = tf.placeholder(tf.int32, [FLAGS.batch_size, FLAGS.height, FLAGS.width], name='ground_truth') _, logits = PSPNet.PSPNet(x, is_training=False, output_stride=FLAGS.output_stride, pre_trained_model=FLAGS.pretrained_model_path,
saved_prediction_val_gray = './pred/val_gray' saved_prediction_test_color = './pred/test_color' saved_prediction_test_gray = './pred/test_gray' VAL_LIST = input_data.VAL_LIST ANNOTATION_PATH = input_data.ANNOTATION_PATH val_num = 1449 test_num = 1456 if not os.path.exists('./pred'): os.mkdir('./pred') val_data = input_data.read_val_data() test_data = input_data.read_test_data() with tf.name_scope("input"): x = tf.placeholder(tf.float32, [BATCH_SIZE, None, None, 3], name='x_input') y = tf.placeholder(tf.int32, [BATCH_SIZE, None, None], name='ground_truth') logits = deeplab_model.deeplab_v3_plus(x, is_training=False, output_stride=8, pre_trained_model=PRETRAINED_MODEL_PATH) with tf.name_scope('prediction_and_miou'): prediction = tf.argmax(logits, axis=-1, name='predictions')