A simple REACT node.js example

 Create Hello.js REACT script

helloAli.js

'use strict';

//document.write("Hello World!");

const name = 'Ali Riza SARAL';

const element = &lth1&gtHello from {name}&lt/h1>

ReactDOM.render(

  element,

  document.getElementById('root')

);

Create the html that will insert the REACT jscript.

index4.html

&lt!doctype html&gt

&lthtml&gt

&lthead&gt

    &ltmeta charset="utf-8"&gt

    &lttitle&gtHello React!&lt/title&gt

    &ltscript src="https://unpkg.com/react@16/umd/react.development.js"&gt&lt/script&gt

    &ltscript src="https://unpkg.com/react-dom@16/umd/react-dom.development.js"&gt&lt/script&gt

    &ltscript src="https://unpkg.com/babel-standalone@6.26.0/babel.js"&gt&lt/script&gt

&lt/head&gt

&ltbody&gt

    &ltdiv id="root"&gt&lt/div&gt

    &ltscript type="text/babel" src="helloAli.js"&gt&lt/script&gt

&lt/body&gt

&lt/html&gt

Create the server that will display the html.

var url = require('url');

// Create a server

http.createServer( function (request, response) { 

   // Parse the request containing file name

   var pathname = url.parse(request.url).pathname;

  

   // Print the name of the file for which request is made.

   console.log("Request for " + pathname + " received.");

  

   // Read the requested file content from file system

   fs.readFile(pathname.substr(1), function (err, data) {

      if (err) {

         console.log(err);

        

         // HTTP Status: 404 : NOT FOUND

         // Content Type: text/plain

         response.writeHead(404, {'Content-Type': 'text/html'});

      } else {           

         //Page found            

         // HTTP Status: 200 : OK

         // Content Type: text/plain

         response.writeHead(200, {'Content-Type': 'text/html'});  

        

         // Write the content of the file to response body

         response.write(data.toString());                   

      }

     

      // Send the response body

      response.end();

   });  

}).listen(8081);

// Console will print the message

console.log('Server running at http://127.0.0.1:8081/');

Run procedure in the same library of all the above:

1-      Start the server with

Node server

2-      Open a browser

3-      Run the REACT through html

http://127.0.0.1:8081/index4.html

Notes:  You do not have to use the  node.js server given here.  Any server Tomcat, Glassfish will work OK.

   

   

   

A short note on the Agile methods

Agile(Çevik) yöntemler Üzerine Kısa Bir Not

“Business people and developers must work together daily throughout the project.

İş süreçlerinin sahipleri ve yazılımcılar proje boyunca her gün birlikte çalışmalıdırlar”.

Bu madde bazı durumlarda Agile yöntemlerin uygulanmasını imkansız kılabilir.

Eğer müşteri tarafında sorunsuz bir ilgi ve destek yok ise Agile yöntemlerin uygulanması bir kaosa neden olabilir. 

Ayrıca müşteri isteklerinin tutarlı ve süreklilik taşıyan niteliklerde olması projenin başarısını doğrudan etkiler.  “İki tane hava trafik kontrolörünün olduğu yerde 3 tane görüş vardır”.

“Değişen gereksinimler yazılım sürecinin son aşamalarında bile kabul edilmelidir.
Çevik süreçler değişimi müşterinin rekabet avantajı için kullanır.”  Bunu her projede uygulamak mümkün olmayabilir.

“Çalışan yazılım, tercihen kısa zaman aralıkları belirlenerek birkaç haftada ya da birkaç ayda bir düzenli olarak müşteriye sunulmalıdır.”  Bu madde yalnız belirli bir büyüklüğe kadar ya da projenin sonuna doğru geçerli olabilir.

Kullanıcı gereksinimleri(user requirements) bütünlük ve doğruluk arzetmek zorunda değildir.

Kullanıcı gereksinimlerindeki eksikler ya da gözükmeyen yapısal gereksinimler yazılım gereksinimleri(software requirements) ile tamamlanır.  Kullanıcının yapısal teknikleri bilmesi mümkün değildir.  Dolayısıyla kullanıcının rolünü bu kadar abartmak sağlıklı sonuçlar vermeyebilir.

how to convert CPU usage to Movidius in Keras

I used a simple Keras example as a starting point. 

Develop Your First Neural Network in Python With Keras Step-By-Step

https://machinelearningmastery.com/tutorial-first-neural-network-python-keras/

I took the first half of this example:

# MLP for Pima Indians Dataset Serialize to JSON and HDF5

from keras.models import Sequential

from keras.layers import Dense

from keras.models import model_from_json

import numpy

import os

# fix random seed for reproducibility

numpy.random.seed(7)

# load pima indians dataset

dataset = numpy.loadtxt("pima-indians-diabetes.csv", delimiter=",")

# split into input (X) and output (Y) variables

X = dataset[:,0:8]

Y = dataset[:,8]

# create model

model = Sequential()

model.add(Dense(12, input_dim=8, kernel_initializer='uniform', activation='relu'))

model.add(Dense(8, kernel_initializer='uniform', activation='relu'))

model.add(Dense(1, kernel_initializer='uniform', activation='sigmoid'))

# Compile model

model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])

# Fit the model

model.fit(X, Y, epochs=150, batch_size=10, verbose=0)

x=numpy.array([[0.42030057, 0.78084083, 0.76165254, 0.19794683, 0.78010274, 0.24512312, 0.17131911, 0.03071891]])

print(x)

yy=model.predict(x, batch_size=None, verbose=0, steps=None)

print(yy)

and I added prediction stuff:

x=numpy.array([[0.42030057, 0.78084083, 0.76165254, 0.19794683, 0.78010274, 0.24512312, 0.17131911, 0.03071891]])

print(x)

yy=model.predict(x, batch_size=None, verbose=0, steps=None)

print(yy)

the output follows:

runfile('C:/Users/ars/.spyder-py3/ars-test/keras_first_network.py', wdir='C:/Users/ars/.spyder-py3/ars-test')

768/768 [==============================] - 0s 13us/step

acc: 79.30%

[[0.42030057 0.78084083 0.76165254 0.19794683 0.78010274 0.24512312

  0.17131911 0.03071891]]

[[0.00283898]]

This was all done with Anaconda using Keras.  I put this to make a comparison later on.

pimasARS.py is the same as the Keras version.

# MLP for Pima Indians Dataset Serialize to JSON and HDF5

from keras.models import Sequential

from keras.layers import Dense

from keras.models import model_from_json

import numpy

import os

# fix random seed for reproducibility

numpy.random.seed(7)

# load pima indians dataset

dataset = numpy.loadtxt("pima-indians-diabetes.csv", delimiter=",")

# split into input (X) and output (Y) variables

X = dataset[:,0:8]

Y = dataset[:,8]

# create model

model = Sequential()

model.add(Dense(12, input_dim=8, kernel_initializer='uniform', activation='relu'))

model.add(Dense(8, kernel_initializer='uniform', activation='relu'))

model.add(Dense(1, kernel_initializer='uniform', activation='sigmoid'))

# Compile model

model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])

# Fit the model

model.fit(X, Y, epochs=150, batch_size=10, verbose=0)

# evaluate the model

scores = model.evaluate(X, Y, verbose=0)

print("%s: %.2f%%" % (model.metrics_names[1], scores[1]*100))

but it saves the model to Keras model files JSON and HDF5.

# serialize model to JSON

model_json = model.to_json()

with open("model.json", "w") as json_file:

    json_file.write(model_json)

# serialize weights to HDF5

model.save_weights("model.h5")

print("Saved model to disk")

Movidius needs to work with TensorFlow  graph files.  So I used

Convert-pimas file from

How to run Keras model on Movidius neural compute stick

https://github.com/Tony607/keras_mnist

convert-pimas.py

from keras.models import model_from_json

from keras import backend as K

import tensorflow as tf

from keras.models import model_from_json

from keras import backend as K

import tensorflow as tf

model_file = "model.json"

weights_file = "model.h5"

with open(model_file, "r") as file:

    config = file.read()

K.set_learning_phase(0)

model = model_from_json(config)

model.load_weights(weights_file)

saver = tf.train.Saver()

sess = K.get_session()

saver.save(sess, "./TF_Model/tf_model")

fw = tf.summary.FileWriter('logs', sess.graph)

fw.close()

model_file = "model.json"

weights_file = "model.h5"

with open(model_file, "r") as file:

    config = file.read()

K.set_learning_phase(0)

model = model_from_json(config)

model.load_weights(weights_file)

saver = tf.train.Saver()

sess = K.get_session()

saver.save(sess, "./TF_Model/tf_model")

fw = tf.summary.FileWriter('logs', sess.graph)

fw.close()

if you notice, this program does not produce a TensorFlow

graph file.  This job is done with the statement below

which you can find in the Makefile.

.PHONY: compile

compile: weights

                test -f graph || ${NCCOMPILE} -s 12 ${MODEL_FILENAME} ${INPUT_NODE_FLAG} ${OUTPUT_NODE_FLAG}

mvNCCompile -s 12 TF_Model/tf_model.meta -in=dense_1_input -on=dense_3/Sigmoid

-in and –on parameters can be found in the logs/events.out.tfevents.1549033801.ars files

After this we have to prepare the Movidius stick, open, load etc.

İn the predict-pimas.py.

#!/usr/bin/env python3.5

# [NCSDK2 API](https://movidius.github.io/ncsdk/ncapi/ncapi2/py_api/readme.html)

from mvnc import mvncapi as mvnc

from keras import backend as K

import numpy

import cv2

# Using NCS Predict

# set the logging level for the NC API

# mvnc.global_set_option(mvnc.GlobalOption.RW_LOG_LEVEL, 0)

# get a list of names for all the devices plugged into the system

devices = mvnc.enumerate_devices()

if len(devices) == 0:

    print('No devices found')

    quit()

# get the first NCS device by its name.  For this program we will always open the first NCS device.

dev = mvnc.Device(devices[0])

# try to open the device.  this will throw an exception if someone else has it open already

try:

    dev.open()

except:

    print("Error - Could not open NCS device.")

    quit()

Once Movidius stick is ready we have to load the graph to it

# Read a compiled network graph from file (set the graph_filepath correctly for your graph file)

with open("graph", mode='rb') as f:

    graphFileBuff = f.read()

graph = mvnc.Graph('graph1')

then we have to put in the test data (the same as the complete Keras version)

# Allocate the graph on the device and create input and output Fifos

in_fifo, out_fifo = graph.allocate_with_fifos(dev, graphFileBuff)

testInput=numpy.array([[0.42030057, 0.78084083, 0.76165254, 0.19794683, 0.78010274, 0.24512312, 0.17131911, 0.03071891]])

# Write the input to the input_fifo buffer and queue an inference in one call

#graph.queue_inference_with_fifo_elem(in_fifo, out_fifo, testInput.astype('float32'), 'user object')

graph.queue_inference_with_fifo_elem(in_fifo, out_fifo, testInput.astype('float32'),'object1')

read the output and print

# Read the result to the output Fifo

output, userobj = out_fifo.read_elem()

# Deallocate and destroy the fifo and graph handles, close the device, and destroy the device handle

try:

    in_fifo.destroy()

    out_fifo.destroy()

    graph.destroy()

    dev.close()

    dev.destroy()

except:

    print("Error - could not close/destroy Graph/NCS device.")

    quit()

#print("NCS \r\n", output, '\r\nPredicted:',output.argmax())

print(output)

The output :

arsaral@ars:~/ncsdk/ncappzoo/apps/pimas$ python3  predict-pimas.py

Using TensorFlow backend.

/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/ops.py:936: DeprecationWarning: builtin type EagerTensor has no __module__ attribute

  EagerTensor = c_api.TFE_Py_InitEagerTensor(_EagerTensorBase)

/usr/local/lib/python3.5/dist-packages/tensorflow/python/util/tf_inspect.py:75: DeprecationWarning: inspect.getargspec() is deprecated, use inspect.signature() instead

  return _inspect.getargspec(target)

/usr/local/lib/python3.5/dist-packages/mvnc/mvncapi.py:416: DeprecationWarning: The binary mode of fromstring is deprecated, as it behaves surprisingly on unicode inputs. Use frombuffer instead

  tensor = numpy.fromstring(tensor.raw, dtype=numpy.float32)

[0.00512314]

arsaral@ars:~/ncsdk/ncappzoo/apps/pimas$

A Keras only test output: [0.00283898]

A Movidius test output:  [0.00512314]