For each sample, I have a 2D array that is NOT an image that I would like to do inference on via tensorflow serving. In the past, I have been able to deploy tensorflow serving successfully thanks to the answer to this post which uses the following serving_input_receiver_fn:
HEIGHT = 199
WIDTH = 199
CHANNELS = 1
def serving_input_receiver_fn():
def decode_and_resize(image_str_tensor):
"""Decodes jpeg string, resizes it and returns a uint8 tensor."""
image = tf.image.decode_jpeg(image_str_tensor, channels=CHANNELS)
image = tf.expand_dims(image, 0)
image = tf.image.resize_bilinear(
image, [HEIGHT, WIDTH], align_corners=False)
image = tf.squeeze(image, squeeze_dims=[0])
image = tf.cast(image, dtype=tf.uint8)
return image
# Optional; currently necessary for batch prediction.
key_input = tf.placeholder(tf.string, shape=[None])
key_output = tf.identity(key_input)
input_ph = tf.placeholder(tf.string, shape=[None], name='image_binary')
images_tensor = tf.map_fn(
decode_and_resize, input_ph, back_prop=False, dtype=tf.uint8)
images_tensor = tf.image.convert_image_dtype(images_tensor, dtype=tf.float32)
return tf.estimator.export.ServingInputReceiver(
{'images': images_tensor},
{'bytes': input_ph})
However, for nonimage arrays, the following becomes unclear:
- How to decode the encoded string tensor. I took a look at
tf.io.decode_image, but it doesn't seem to preserve the 2D array dimension. - How to encode the array. In the case of images, I encoded the image data itself by
base64.b64encode(img_data). For general 2D arrays, how should I encode them?
In short what is the way to generalize the linked post's answer to the non image array case?
