SNPE Tutorial (tensorflow)

import os, argparse

import tensorflow as tf

# The original freeze_graph function

# from tensorflow.python.tools.freeze_graph import freeze_graph

dir = os.path.dirname(os.path.realpath(__file__))

def freeze_graph(model_dir, output_node_names):

    """Extract the sub graph defined by the output nodes and convert

    all its variables into constant

    Args:

        model_dir: the root folder containing the checkpoint state file

        output_node_names: a string, containing all the output node's names,

                            comma separated

    """

    if not tf.gfile.Exists(model_dir):

        raise AssertionError(

            "Export directory doesn't exists. Please specify an export "

            "directory: %s" % model_dir)

    if not output_node_names:

        print("You need to supply the name of a node to --output_node_names.")

        return -1

    # We retrieve our checkpoint fullpath

    checkpoint = tf.train.get_checkpoint_state(model_dir)

    input_checkpoint = checkpoint.model_checkpoint_path

    # We precise the file fullname of our freezed graph

    absolute_model_dir = "/".join(input_checkpoint.split('/')[:-1])

    output_graph = absolute_model_dir + "/frozen_model.pb"

    # We clear devices to allow TensorFlow to control on which device it will load operations

    clear_devices = True

    # We start a session using a temporary fresh Graph

    with tf.Session(graph=tf.Graph()) as sess:

        # We import the meta graph in the current default Graph

        saver = tf.train.import_meta_graph(input_checkpoint + '.meta', clear_devices=clear_devices)

        # We restore the weights

        saver.restore(sess, input_checkpoint)

        # We use a built-in TF helper to export variables to constants

        output_graph_def = tf.graph_util.convert_variables_to_constants(

            sess,  # The session is used to retrieve the weights

            tf.get_default_graph().as_graph_def(),  # The graph_def is used to retrieve the nodes

            output_node_names.split(",")  # The output node names are used to select the usefull nodes

        )

        # Finally we serialize and dump the output graph to the filesystem

        with tf.gfile.GFile(output_graph, "wb") as f:

            f.write(output_graph_def.SerializeToString())

        print("%d ops in the final graph." % len(output_graph_def.node))

    return output_graph_def

if __name__ == '__main__':

    parser = argparse.ArgumentParser()

    parser.add_argument("--model_dir", type=str, default="", help="Model folder to export")

    parser.add_argument("--output_node_names", type=str, default="",

                        help="The name of the output nodes, comma separated.")

    args = parser.parse_args()

    freeze_graph(args.model_dir, args.output_node_names)

python freeze_graph.py --model_dir=./Y_log_transpose --output_node_names='edsr/Conv2d_transepose/BiasAdd'

self.output = self.generator(self.LR, self.filters)

print(self.LR) # input node name

print(self.output) # output node name

from __future__ import print_function

from utils import *

from imresize import imresize

from scipy.misc import imread

import numpy as np

def CalcuPSNR(target, ref):

    target = np.clip(target, 0, 255.0)

    squared_error = np.square(target.astype('uint8').astype('float32') - ref.astype('uint8').astype('float32'))

    psnr = 10 * np.log10(255.0 * 255.0 / np.mean(squared_error))

    return psnr

def rgb2gray(img):

    y = np.sum(img * np.reshape([65.481, 128.553, 24.966], [1, 1, 3]) / 255.0, axis=2) + 16

    return y.astype('uint8')

graph = tf.Graph()

sess = tf.InteractiveSession(graph=graph)

with tf.gfile.FastGFile('./Y_log/frozen_model.pb', 'rb') as f:

    graph_def = tf.GraphDef()

    graph_def.ParseFromString(f.read())

inputImage = tf.placeholder(tf.float32, [1,256,256,1], name = "inputimage")

groudtruth = tf.placeholder(tf.float32, [1,512,512,1], name = "groundtruth")

model = tf.import_graph_def(graph_def, input_map={'random_shuffle_queue_DequeueMany':inputImage}, return_elements=['edsr/DepthToSpace'])

output = model[0].outputs[0]

print('------------------')

lr_img = imread('datasets/val/LR/Set5_LR_bicubic/X2/babyx2.png')

ori_img = imread('datasets/val/HR/Set5_HR/baby.png')

lr_gray = rgb2gray(lr_img)

ori_gray = rgb2gray(ori_img)

bic_gray = imresize(lr_gray,output_shape=(512, 512))

bic_psnr = CalcuPSNR(bic_gray, ori_gray)

lr_gray = (lr_gray / 255.0) * 2 -1

lr_gray = np.expand_dims(lr_gray, axis=0)

lr_gray = np.expand_dims(lr_gray, axis=3)

recon = sess.run(output, feed_dict={inputImage:lr_gray})

recon_img = (recon + 1) * 255 * 0.5

sr_psnr= CalcuPSNR(recon_img[0,:,:,0], ori_gray)

print(bic_psnr)

print(sr_psnr)

usage: snpe-tensorflow-to-dlc [-h] --graph GRAPH -i INPUT_NAME INPUT_DIM

                              --out_node OUT_NODE [--dlc DLC]

                              [--model_version MODEL_VERSION]

                              [--in_type {default,image}]

                              [--allow_unconsumed_nodes] [--verbose]

snpe-tensorflow-to-dlc --graph ./Y_log_transpose/frozen_model.pb -i Placeholder_2 1,256,256,1 --out_node edsr/Conv2d_transpose/BiasAdd --allow_unconsumed_nodes

DESCRIPTION:

------------

Example application demonstrating how to load and execute a neural network

using the SNPE C++ API.

REQUIRED ARGUMENTS:

-------------------

  --container  <FILE>   Path to the DL container containing the network.

  --input_list <FILE>   Path to a file listing the inputs for the network.

OPTIONAL ARGUMENTS:

-------------------

  --use_fxp_cpu         Use the CPU fixed point runtime for SNPE.

  --use_gpu             Use the GPU runtime for SNPE.

  --use_dsp             Use the DSP fixed point runtime for SNPE.

  --use_aip             Use the AIP fixed point runtime for SNPE.

  --debug               Specifies that output from all layers of the network

                        will be saved.

  --output_dir <DIR>    The directory to save output to. Defaults to ./output

  --storage_dir <DIR>   The directory to store SNPE metadata files

  --encoding_type <VAL> Specifies the encoding type of input file. Valid settings are "nv21".

                        Cannot be combined with --userbuffer*.

  --userbuffer_float    [EXPERIMENTAL] Specifies to use userbuffer for inference, and the input type is float.

                        Cannot be combined with --encoding_type.

  --userbuffer_tf8      [EXPERIMENTAL] Specifies to use userbuffer for inference, and the input type is tf8exact0.

                        Cannot be combined with --encoding_type.

  --perf_profile <VAL>  Specifies perf profile to set. Valid settings are "system_settings" , "power_saver" , "balanced" ,

                        "default" , "high_performance" , "sustained_high_performance" , and "burst".

                        NOTE: "balanced" and "default" are the same.  "default" is being deprecated in the future.

  --profiling_level <VAL> Specifies the profiling level.  Valid settings are "off", "basic" and "detailed".

                          Default is detailed.

                          Basic profiling only applies to DSP runtime.

  --enable_cpu_fallback Enables cpu fallback functionality. Defaults to disable mode.

  --input_name <INPUT_NAME> Specifies the name of input for which dimensions are specified.

  --input_dimensions <INPUT_DIM>  Specifies new dimensions for input whose name is specified in input_name. e.g. "1,224,224,3".

                        For multiple inputs, specify --input_name and --input_dimensions multiple times.

  --gpu_mode <VAL>      Specifies gpu operation mode. Valid settings are "default", "float16".

                        default = float32 math and float16 storage (equiv. use_gpu arg).

                        float16 = float16 math and float16 storage.

  --help                 Show this help message.

  --version              Show SNPE Version Number.

test_data/input_baby.raw

test_data/input_car.raw

test_data/input_balloon.raw

test_data/input_bus.raw

...

lr_gray = imread('test_data/babyx2_gray.png')

lr_gray = (lr_gray / 255.0) * 2 -1

lr_gray = np.expand_dims(lr_gray, axis=0)

lr_gray = np.expand_dims(lr_gray, axis=3)

lr_gray = lr_gray.astype(np.float32)

lr_gray.tofile('test_data/input_baby.raw')  # NHWC order

snpe-net-run --container ./Y_log_transpose/frozen_model.dlc --input_list img_list.txt

out_raw = np.fromfile('output_transpose/Result_0/edsr/Conv_2d_transpose/BiasAdd:0.raw', dtype=np.float32)

out_raw = out_raw.reshape((1, 512,512,1))

out_raw = np.transpose(out_raw, [0, 3, 1, 2])

out_raw = (out_raw + 1) * 255.0 * 0.5

out_raw = (out_raw[0, 0, :, :]).clip(0, 255)

sr_snpe_psnr= CalcuPSNR(out_raw, ori_gray)

export SNPE_TARGET_ARCH=arm-android-gcc4.9

export SNPE_TARGET_STL=libgnustl_shared.so

adb shell "mkdir -p /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin"

adb shell "mkdir -p /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib"

adb shell "mkdir -p /data/local/tmp/snpeexample/dsp/lib"

# 아래는 현재 경로가 snpe-1.17.0 라고 생각하고 진행한다

adb push lib/$SNPE_TARGET_ARCH/$SNPE_TARGET_STL /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

adb push lib/$SNPE_TARGET_ARCH/libsymphony-cpu.so /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

adb push lib/$SNPE_TARGET_ARCH/libsymphonypower.so /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

adb push lib/dsp/libsnpe_dsp_skel.so /data/local/tmp/snpeexample/dsp/lib

adb push lib/dsp/libsnpe_dsp_domains_skel.so /data/local/tmp/snpeexample/dsp/lib

adb push lib/dsp/libsnpe_dsp_v65_domains_v2_skel.so /data/local/tmp/snpeexample/dsp/lib

adb push lib/$SNPE_TARGET_ARCH/libsnpe_adsp.so /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

adb push lib/$SNPE_TARGET_ARCH/libsnpe_dsp_domains.so /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

adb push lib/$SNPE_TARGET_ARCH/libSNPE.so /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

adb push bin/$SNPE_TARGET_ARCH/snpe-net-run /data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin

adb devices

adb push test_data/input_baby.raw /data/local/tmp/edsr

adb push img_list.txt /data/local/edsr

adb push frozen_model.dlc /data/local/edsr

adb shell

export SNPE_TARGET_ARCH=arm-android-gcc4.9

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/lib

export PATH=$PATH:/data/local/tmp/snpeexample/$SNPE_TARGET_ARCH/bin

cd /data/local/tmp/edsr

snpe-net-run --container frozen_model.dlc --input_list img_list.txt --use_gpu

exit

adb pull /data/local/tmp/edsr/output output_android_gpu

export SNPE_ROOT=/home/kaist/Desktop/snpe-1.17.0

python snpe_bench.py -c config.json -a

usage: snpe_bench.py [-h] -c CONFIG_FILE [-o OUTPUT_BASE_DIR_OVERRIDE]

                     [-v DEVICE_ID_OVERRIDE] [-a] [-t DEVICE_OS_TYPE_OVERRIDE]

                     [-r HOST_NAME] [-d] [-s SLEEP] [-b USERBUFFER_MODE] [-p PERFPROFILE]