Guidance for Compiling TensorFlow-Slim Networks

If you are compiling a TensorFlow-Slim network for use with the Intel® Movidius™ Neural Compute SDK (Intel® Movidius™ NCSDK) and Neural Compute API, you can follow these steps.

The code below shows how you can save a TensorFlow™ session with graph and checkpoint information.

import numpy as np
import tensorflow as tf

from tensorflow.contrib.slim.nets import inception

slim = tf.contrib.slim

def run(name, image_size, num_classes):
    with tf.Graph().as_default():
      image = tf.placeholder("float", [1, image_size, image_size, 3], name="input")
      with slim.arg_scope(inception.inception_v1_arg_scope()):
          logits, _ = inception.inception_v1(image, num_classes, is_training=False, spatial_squeeze=False)
      probabilities = tf.nn.softmax(logits)
      init_fn = slim.assign_from_checkpoint_fn('inception_v1.ckpt', slim.get_model_variables('InceptionV1'))

    with tf.Session() as sess:
        saver = tf.train.Saver(tf.global_variables()), "output/"+name)

run('inception-v1', 224, 1001)

The is_training=False parameter is important. This will leave out training-only layers (which aren’t supported by the NCSDK) from the network.

Next, use the NCSDK mvNCCompile tool to compile the saved session from the above code sample for use with the NCSDK and Neural Compute API:

mvNCCompile output/inception-v1.meta -in=input -on=InceptionV1/Logits/Predictions/Reshape_1 -s 12