Machine Learn on device performance (Tensor flow way)

Quantize Neural Networks

• curl http://download.tensorflow.org/models/image/imagenet/inception-2015-12-05.tgz -o /tmp/inceptionv3.tgz
  tar xzf /tmp/inceptionv3.tgz -C /tmp/
  bazel build tensorflow/tools/quantization/tools:quantize_graph
  bazel-bin/tensorflow/tools/quantization/tools/quantize_graph \
    --input=/tmp/classify_image_graph_def.pb \
    --output_node_names="softmax" --output=/tmp/quantized_graph.pb \
    --mode=eightbit

• 
  

• bazel build tensorflow/examples/label_image:label_image
  bazel-bin/tensorflow/examples/label_image/label_image \
  --image=<input-image> \
  --graph=/tmp/quantized_graph.pb \
  --labels=/tmp/imagenet_synset_to_human_label_map.txt \
  --input_width=299 \
  --input_height=299 \
  --input_mean=128 \
  --input_std=128 \
  --input_layer="Mul:0" \
  --output_layer="softmax:0"
  

• Song Han's code books
• https://github.com/google/gemmlowp
•