Sppnet input image preprocessing

[tensorflowt]

x_train = tf.image.resize_images(x_train,
[int(size_cluster_keys[it%len(size_cluster_keys)][1]/2),
int(size_cluster_keys[it%len(size_cluster_keys)][0]/2)],
method=1, align_corners=False)

Shouldn't it be like this：
x_train = tf.image.resize_images(x_train,
[int(size_cluster_keys[it%len(size_cluster_keys)][0]/2),
int(size_cluster_keys[it%len(size_cluster_keys)][1]/2)],
method=1, align_corners=False)

[peace195]

Because PIL.Image.size returns (width, height) while tf.image.resize needs (new_height, new_width) parameters.

You can read more about it here: https://www.tensorflow.org/api_docs/python/tf/image/resize_images and https://pillow.readthedocs.io/en/3.1.x/reference/Image.html