Please make sure that this is a feature request. As per our GitHub Policy, we only address code/doc bugs, performance issues, feature requests and build/installation issues on GitHub. tag:feature_template

System information

• TensorFlow version (you are using): tf 2.2.0
• Are you willing to contribute it (Yes/No): Yes

Describe the feature and the current behavior/state.

Currently, the tf.image.resize_with_crop_or_pad ops pads with 0. It would be helpful to be able to give another value because in some application black is not the right color for padding (it messes with real data).

Will this change the current api? How?

Adding a kwarg padding_color Union[int, iterable] with:

• int, a value to be used on all channels (currently 0)
• iterable: with len(iterable) == number of channel to describe any rgb color

Example of how to achieve the expected result currently:

        output_tensor = tf.image.resize_with_crop_or_pad(input_tensor, 450, 450)
        padding = (1 - tf.image.resize_with_crop_or_pad(tf.ones_like(input_tensor), 450, 450)) * tf.constant(
            [255, 0, 255], dtype=tf.uint8
        )
        output_tensor += padding

Who will benefit with this feature?

Users who need to pad images with another color because black is not suitable for their dataset

Any Other info.