Implementation of contrast() seems wrong

[EIFY]

L209 of autoaugment.py

tpu/models/official/efficientnet/autoaugment.py

Lines 205 to 209 in c3186a4

	# Compute the grayscale histogram, then compute the mean pixel value,
	# and create a constant image size of that value. Use that as the
	# blending degenerate target of the original image.
	hist = tf.histogram_fixed_width(degenerate, [0, 255], nbins=256)
	mean = tf.reduce_sum(tf.cast(hist, tf.float32)) / 256.0

and L278 of autoaugment_utils.py of the implementation of contrast()

tpu/models/official/detection/utils/autoaugment_utils.py

Lines 274 to 278 in c3186a4

	# Compute the grayscale histogram, then compute the mean pixel value,
	# and create a constant image size of that value. Use that as the
	# blending degenerate target of the original image.
	hist = tf.histogram_fixed_width(degenerate, [0, 255], nbins=256)
	mean = tf.reduce_sum(tf.cast(hist, tf.float32)) / 256.0

seem wrong. mean is supposed to be the mean pixel value, but as it is it's just summing over the histogram (therefore equal to height * width), divided by 256. google-research/big_vision and tensorflow/models have the same bug so ideally should be fixed in the same way. See google-research/big_vision#109 for details including the manual tests I did.

[EIFY]

Note: 4-year-old PR #764 fixes the implementation in efficientnet/