Change the image size as output about semantic_segmentation

[martsalz]

def predict_images(test_data_path, print_speed=False):
num_classes = 2
image_shape = (557, 1618)

If the shape is changed in the main.py in the function predict_images, the following exception is thrown:

tensorflow.python.framework.errors_impl.InvalidArgumentError: Incompatible shapes: [1,2,36,102] vs. [1,2,35,102]

[Shaiken]

i had sma question