Either through https://www.opencv.org or direct link: https://sourceforge.net/projects/opencvlibrary/files/4.0.0/opencv-4.0.0-vc14_vc15.exe/download

Unzip to to c:\opencv40

Your directory should look like this:

1. From Windows start menu, run “Edit the System Environment Variables”
2. Click on Environment Variables and click New. Define a new environment variable named OPENCV_DIR and set it to the location of the OpenCV directory (usually
   c:\Opencv40\build where 40 means OpenCV 4.0). Exit the dialog box.

3. Install Visual Studio 2017
4. Create a simple console application (ConsoleApplication1).
5. Go to Project | ConsoleApplication1 Properties… and select the configuration All Configurations. Platform should be x64. Select C/C++, General and click on the Additional Include Directories, and select Edit.
6. Add the directory $(OPENCV_DIR)\include and ensure it is properly evaluated

7. Do the same in Linker, General and Additional Library Directories:
$(OPENCV_DIR)\x64\vc15\lib

8. In Debugging, add the command: PATH=$(OPENCV_DIR)\x64\vc15\bin;%PATH%

9. Add the entry $(OPENCV_DIR)\x64\vc15\lib\opencv_world400d.lib to the Debug configuration

10. Repeat the same with the Release dll: Add the entry $(OPENCV_DIR)\x64\vc15\lib\opencv_world400.lib to the Release configuration (notice that the ending is 400.dll and not 400d.dll)

11. Create an image named test.png in ConsoleApplication directory
12. Write the following code
13. Run and see that it works

// ConsoleApplication1.cpp : This file contains the 'main' function. Program execution begins and ends there.
//

#include "pch.h"
#include <iostream>
#include <opencv2/opencv.hpp>

using namespace cv;
int main()
{
  Mat img=cv::imread("test.png");
  cv::imshow("Test", img);
  cv::waitKey();
}

#include "pch.h"
#include <iostream>
#include <opencv2/opencv.hpp>

int main()
{
    cv::VideoCapture cap(0); // open the default camera
    if (!cap.isOpened())  // check if we succeeded
        return -1;

    cv::Mat edges;
    cv::namedWindow("edges", 1);
    for (;;)
    {
 		cv::Mat frame;
        cap >> frame; // get a new frame from camera
        cv::cvtColor(frame, edges, cv::COLOR_BGR2GRAY);
        cv::GaussianBlur(edges, edges, cv::Size(7, 7), 1.5, 1.5);
        cv::Canny(edges, edges, 0, 30, 3);
        cv::imshow("edges", edges);
        if (cv::waitKey(30) >= 0) break;
    }
    // the camera will be deinitialized automatically in VideoCapture destructor
    return 0;
}

import numpy as np
import cv2
cap = cv2.VideoCapture(0)

while(True):
    # Capture frame-by-frame
    ret, frame = cap.read()
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY):
    cv2.imshow('frame',gray)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
cap.release()
cv2.destroyAllWindows()

1. Install Oracle VM VirtualBox latest version 5.2.20 from https://www.virtualbox.org/wiki/Downloads. Username: sightec, password: sightex

2. Generate a 64 bit Ubuntu machine (20GB HD, 8GB RAM) with clipboard sharing

3. Install Ubuntu 18.04 LTS from https://www.ubuntu.com/#download

4. Upgrad and update all packages to the latest version

pip sudo apt-get update
sudo apt-get upgrade

5. Run Software Updater (Ubuntu)

6. Install Guest additions on Ubuntu (From VirtualBox, Devices|Insert Guest Additions CD Image…) and reboot. This will enable clipboard sharing and file sharing between the host and the virtual machine

7. In Ubuntu, set up the tool chain (compilers and libraries, see Required Packages in https://docs.opencv.org/3.4/d7/d9f/tutorial_linux_install.html)

sudo apt-get install build-essential

sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev

sudo apt-get install libtbb2 libtbb-dev libjpeg-dev libpng-dev libtiff-dev libjasper-dev libdc1394-22-dev

8. Set Python and related libraries
1. To get the most out of python, install the suite: ipython (interactive python development environment with shell and code completion), numpy (numerical librarie) and matplotlib (graphs and MATLAB feel)

sudo apt-get install python-dev python-numpy python-dbg ipython python-matplotlib

9. Some additional libraries for OpenCV:

sudo apt-get install libgtk-3-dev

10. Get the latest version of OpenCV as source code (zipped file) from https://opencv.org/releases.html

unzip Downloads/opencv-3.4.3.zip

11. Create a temporary directory to compile OpenCV:

mkdir -p ~/temp_opencv_compile/build

12. Configure cmake-gui

◦ If cmake-gui is not insalled, sudo apt install cmake-qt-gui
◦ run cmake-gui from the command line
◦ Make sure to configure cmake-gui with CodeBlocks – Unix Makefiles so we can later edit/view the source code in Code::Blocks C/C++ IDE
◦ Where is the source code: /home/sightec/opencv-3.4.3.
◦ Where to build binaries: /home/sightec/temp_opencv_compile/build.
◦ Click configure. Use default compile settings.
◦ Ensure to select BUILD_WITH_DEBUG_INFO, CMAKE_CONFIGURATION_TYPES only Debug (make sure to delete the Release configurations), CMAKE_BUILD_TYPE=Debug, ENABLE_GNU_STL_DEBUG (support for standard template library structures debug such as vectors and maps). This will enable debugging OpenCV C/C++ code.
◦ Click Generate
◦ View the results and ensure support for the modules you wish. If, for example, you want Cuda support and the configuration result shows no Cuda support, install Cuda using apt-get and re-run cmake-gui
◦ Exit cmake-gui

13. Make OpenCV

make -j4 (or simply make)

14. Install OpenCV

sudo make install

(note: you must sudo)

Scroll back and ensure the the Debug version was installed (and not the Release). It should read: — Install configuration: “Debug”

15. Test OpenCV from Python

Code:

import cv2
from pylab import *
n = randn(100,100)
thr, n = cv2.threshold(n, 0.5, 1, cv2.THRESH_BINARY)
ion()
imshow(n)

16. Note: I’ve covered Code::Blocks (C/C++ compiled/debugger); it’s possible to use Eclipse as well.

17. Install Code::Blocks (http://codeblocks.org/): sudo apt-get install codeblocks

18. Launch codeblocks and open the existing project from ~/temp_opencv_compile/build named OpenCV.cbp

19. From Code::Blocks main bar, select the click on the drop down box and select opencv_features2d.

20. Click on Project||Set programs’ arguments| Set and type in the host application text box
/usr/bin/python. Also check the Run host in terminal checkbox.

21. Add breakpoints in the C/C++ code as you see fit.

22. In Code::Blocks, click on Debug/Continue button. This will do a fast compilation (pre-compiled headers check) and launch a terminal.

23. You will be presented with a terminal and Python running inside. Issue the following (example):

import cv2

img =cv2.imread(‘/home/sightec/index.png’)

orb=cv2.ORB_create()

orb.detect(img)

24. You might experience throws from OpenCL so either disable them in the compilation phase of OpenCV or continue and keep debugging (it is only thrown once).

25. Add watches/breakpoints as you see fit and continue.

26. If you have ipython installed (highly recommended) you can run it from the debugger instead of the python console as follows

1. Host application: /bin/bash
2. Program arguments: /usr/bin/ipython

1. To get MATLAB-like functionality, install the suite:

a) Ipython, an interactive python development environment with shell and code completion, see https://ipython.org/
b) Numpy, a numerical library, see http://www.numpy.org/
c) matplotlib, a powerful package for plotting, https://matplotlib.org/

sudo apt install ipython python-matplotlib/

2. If you’re also interested in image processing, install OpenCV and bindings. Note that there are two ways to install OpenCV; this method will install compiled OpenCV with Python bindings. If you wish to install the source code so you can edit OpenCV, see https://opencv.org/

sudo apt install python-opencv

3. For additional MATLAB toolboxes such as signal processing, special functions, integration and interpolation, optimization, sparse matrices and more, install SciPy, https://www.scipy.org/:

sudo apt install python-scipy

4. For machine learning capabilities, install scikit-learn, https://scikit-learn.org/stable/

sudo apt install python-sklearn

5. For additional Python packages no covered in this document, consult with PyPI, the Python Package Index, at https://pypi.org/

Ipython is an enhanced interactive python environment with command completion and syntax highlghting. Some useful commands to get you started are: ls, cd, help, reset and run.

Start Ipython and import MATLAB functionality

$ ipython –pylab

Alternatively, you can start python/ipython and then import pylab:

$ ipython

from pylab import *

Pylab combines the functionality of Numpy, Scipy, Ipython and matplotlib.

See that you can plot a simple graph:

t=arange(0, 100, 0.01)
plot(t, sin(t))
title('Sin(t)')
xlabel('time [sec]')
grid()
show()

By default, graphs are not shown until you issue the command show(). You can override this by calling the command ion() or ioff().

Matplotlib supports a wide range of graphs. See this gallery (including code) for its capabilities, see https://matplotlib.org/gallery/index.html

As a general rule, use only Numpy’s ndarray data structure and not the mat data structure. The mat data structure is not supported by all packages (whereas ndarray is). This is the building block for all numerical operations and packages on Python. See https://docs.scipy.org/doc/numpy-1.15.0/user/basics.creation.html. If you’re using pylab, you can drop the leading np.. If you want to manually use Numpy, issue import numpy as np.

Examples:
arange(10)
zeros((2,2,3))

See the link above. To use, import scipy.

Scikit-learn (or sklearn for short), supports a great deal of machine learning methodologies and algorithms including: classification, regression, clustering, and dimensionality reduction. It also provides analysis and validation tools.
The scikit-learn.org has a great deal of information available. Start with the tutorials https://scikit-learn.org/stable/tutorial/index.html and move on to explore the examples in https://scikit-learn.org/stable/auto_examples/index.html.
To use sklearn: import sklearn

1. Install the latest Python 2 release (notice, 2, not 3), from https://www.python.org/downloads/windows/. Please make sure you download the 64 bit install version (quite important when doing machine learning with Scikit-learn), named Windows x86-64 MSI installer. Install on your PC for all users. Install Python to folder c:\python27
2. Optional: install an enhanced command console ConEmu from https://conemu.github.io/. Alternatively, you can use a regular command prompt.
3. Additional packages, see PYPI
4. Optional: install notepad++, https://notepad-plus-plus.org/download/v7.6.html
5. Install NumPy, Matplotlib and IPython. Open a command console, either by start|Run|Cmd or by running ConEmu. Change directory to c:\python27\scripts and install additional packages:

cd \python27\scripts

pip install numpy matplotlib ipython scipy

1. Open ConEmu and and change directory to c:\python\scripts, run ipytho
ipython

a. Exercise: set ConEmu to run IPython in a console on startup

2. Test ipython, matplotlib and numpy.

a. Issue from pylab import *
b. Test command completion, start writing pl and press Tab.
c. Test console commands (type ls)
d. Show a graph:

t=arange(0, 100, 0.01)
plot(t, sin(t))
title('Sin(t)')
xlabel('time [sec]')
grid()
show()
# ion()
# ioff()
close('all') – close all open graphs
grid()
savefig(fname)

Matplotlib supports a wide range of graphs, see https://matplotlib.org/gallery/index.html

Character completion

Up/down arrows for last commands

Useful commands:

ls
run
help
reset
cd
pwd
del

Using Python as a calculator

Using the last answer (_)

Notice integer division vs. floating point division: 2/3, 2.0/3, 2//3, 2.0//3

path=%PATH%;c:\python27/span>

There are three several ways to import (load) a library:

1. Import with context (usage: math.sin()):
import math
2. Import the entire library (usage: sin()):
from math import *
3. Import a specific function (usage: sin()):
from math import sin
4. Import a specific function, renamed (usage: sine()):
from math import sin as sine

Exploring a package:

1. Character completion
2. dir(library)

Matlab-like functionality package:
from pylab import *

Int and long, sys.maxint, 2**2000

Absolute precision vs. floating point precision: 3*(1./3+2.2)

Other bases: 0xab, int(‘100’, 5)

Float and complex numbers: 1j, 100.23, a=complex(10,2), a.real, a.imag

Booleans: 2*3>5, True, False

Bitwise operators: 2<<5

Strings: ‘shai’, ‘shai\’s’, “shai”, “shai’s”, r”shai’s”, “””docstring”””

String accessing/indexing/slicing

Lists: [1, ‘shai’], [‘shai’]+[‘shai’], [[1, 2], [3, 4]], append, count, extend, remove, etc.

Tuples, tuple unpacking

Dictionary: d=dict(), d[‘shai’]=2, dict(((1, 2), (3, 4))), clear, get, items, keys, etc.

arrays

Ndarray: https://docs.scipy.org/doc/numpy-1.13.0/reference/routines.array-creation.html

While loop

While condition:

Do_Block

If then else

if condition:

Do_Block

elif:

Do_Block

else:

For loop

for val in iterable:

Do_Block

In [57]: x

Out[57]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

In [58]: for val in x:

…:     print val, val**2

…:

0 0
1 1
2 4
3 9
4 16
5 25
6 36
7 49
8 64
9 81

Zipping arrays

A=[1, 5, 'shai']
B=[3, -1, 1j]
zip(A, B)
[(1, 3), (5, -1), ('shai', 1j)]

For a, b in zip(A, B):
print a, b

Enumerate

In [12]: for i, val in enumerate(A):

…:     print i, val

…:

0 1
1 5
2 shai

While loop

While condition:

Do_Block

Exercise

1. Write a script to sum the digits of a number. Example: 1234 should results 10.
2. Write a script to calculate n!
3. Write a script to calculate sqrt(2) using the bisection method

If then else

if condition:

Do_Block

elif condition2:

Do_Block

else:

Do_Block

For loop

for val in iterable:

Do_Block

In [57]: x

Out[57]: array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

In [58]: for val in x:

…:     print val, val**2

…:

0 0
1 1
2 4
3 9
4 16
5 25
6 36
7 49
8 64
9 81

Enumerate

In [12]: for i, val in enumerate(A):

…:     print i, val

…:

0 1
1 5
2 shai

Functions

In [4]: def f(x):

…:     return(sqrt(1-x**2))

…:

In [5]: f(0)

Out[5]: 1.0

In [6]: f(-1)

Out[6]: 0.0

In [7]: f(1)

Out[7]: 0.0

Functions can also return complex objects (dictionaries, arrays, etc).

Simple files 

Open a file for reading, read the data and close the file:

f = open(filenane)

f = open('pg97.txt')

data = f.read()
f.close()

 
Do it in one line

data = open(filename).read()
 
Read a text file and split it into lines

lines=open(‘noam.py’).readlines()

Excercise

1. Download the file Flatland from http://www.gutenberg.org/cache/epub/97/pg97.txt and analyze the following: how many lines, words and characters in the file. How many unique words are in the file? Generate a new file with every occurrence of Flatland renamed to Spaceland

Hints:
len(), split(), text()

Reading a CSV file 

Use the module urllib to read the file https://www.cia.gov/library/publications/the-world-factbook/rankorder/rawdata_2001.txt and store it locally to file cia.csv

import urllib
f=urllib.urlopen('https://www.cia.gov/library/publications/the-world-factbook/rankorder/rawdata_2001.txt')
data=f.read()
f.close()
fout = open('cia.csv','w')
fout.write(data)
f.close()

Draw a bar chart of the first 10 nations 

top10=open('cia.csv').readlines()[:10]
nations, gdp = [], []
for row in top10:


nations.append(row[7:58].strip())
gdp.append(int(row[59:].strip().replace(',','')))

bar(nations, gdp)
gca().set_xticklabels(nations, rotation=45)

Download QQQ historical data and plot the graph using loadtxt() 

File location: https://finance.yahoo.com/quote/QQQ/history/ , click on Download data and save as QQQ.csv

Skip the first row and use only columns 2, 3, 4 and 5. Plot some graphs:

d = loadtxt(‘QQQ.csv’, skiprows=1, delimiter=’,’, usecols=(2, 3, 4, 5))
for i in range(4): plot(d[:, i])

More elaborate:

d = loadtxt('QQQ.csv', skiprows=1, delimiter=',', converters={0: datestr2num})

for i in range(1, 6):

plot(d[:, i])

gca().set_xticklabels([num2date(x).strftime('%Y-%m-%d') for x in d[:,0]], rotation=45)

Ensure OpenCV is available and check its version:

pip install opencv-python

import cv2
cv2.__version__
‘4.0.0’

Reading an image file

airplane=cv2.imread(‘airplane.jpg’)
Writing image to file: cv2.imwrite(filename, img). The extension defines the image format.

cv2.imshow(‘airplane’, airplane) will display the image, or alternatively, use pylab’s imshow(airplane).

Exercise: write a function to convert jpeg images in the current working directory to PNG images.

Tip: use os.listdir() or
glob.glob() functions to list the directory content

img.shape holds the image shape. If the image is RGB, the size would be 3 dimensional (width, height, 3). To extract just one channel, use the : operator:

R = airplane[:, :, 0]

You can change the colormap:hot(), cool(). See help(colormaps) for more options.

Combining channels can be done similarly.

Another important function is R = airplane[:, :, 0] to convert to gray levels and back.

img = cv2.cvtColor(airplane, cv2.COLOR_BGR2GRAY)
gray()

Need to understand the concept of binding, copy, deep copy, etc.

img = airplane.copy()

How many bytes/bits represent a pixel in an image. Common values cv2.CV_8U, cv2.CV_64F.

See https://docs.opencv.org/master/

Examples:

1. Gaussian blur (highly used).

g = cv2.GaussianBlur(airplane, (51, 51), 3.0)

2. Sobel (derivative)

g = cv2.cvtColor(airplane, cv2.COLOR_BGR2GRAY)
s = cv2.Sobel(g, cv2.CV_64F, 1, 1)
imshow(s)

3. Canny edge detector

edges = cv2.Canny(g,80,180,apertureSize = 3)

4. Hough line transform (or circle transform)

lines = cv2.HoughLinesP(edges,1,np.pi/180,15,minLineLength,maxLineGap)

Final Exercise

Find the image in focus and derive the Z value in directory autofocus.

Tip: Have a look at the image processing documentation of OpenCV

Have a look at cv2.threshold()with OTSU’s method. Highly recommended.

Solution to image conversion:

Option 1: quick and dirty

from glob import glob

for fn in glob(‘*.bmp’):

cv2.imwrite(fn.replace(‘bmp’, ‘jpg’), cv2.imread(fn))

Downside with this option: in case the filename contains the sequence ‘bmp’ it will not work. Example: If the filename is edpbmp32.bmp, the code above will create a file named edpjpg32.bmp (the first bmp occurrence will be replaced).

Option 2: replace the last 3 characters

from glob import glob

for fn in glob(‘*.bmp’):

If you’d like to use os.listdir() instead of glob, use the following:

for fn in os.listdir(‘.’):

if fn.endswith(‘bmp’):

cv2.imwrite(fn.replace(‘bmp’, ‘jpg’), cv2.imread(fn))

In Windows, you probably want this to be case insensitive so consider changing to this:

for fn in os.listdir(‘.’):

if fn.lower().endswith(‘bmp’):

cv2.imwrite(fn[:-3]+’jpg’, cv2.imread(fn))

Solution to the final exercise:

from pylab import *
import cv2, glob

x, y = [], []

for fn in glob.glob('*.bmp'):

    # Sobel is the indicator for focus
    focus = cv2.Sobel(imread(fn), cv2.CV_64F, 1, 1)
    y.append(norm(focus))
    
    # extract the Z from the filename
    start = fn.find('(')+1
    stop = fn.find(')')
    x.append(int(fn[start:stop]))

# plot the focus indicator as a function of z	
plot(x, y, 'ob')

# the value in focus corresponds to argmax
plot(x[argmax(y)], y[argmax(y)], 'or')

# add text on the graph
text(x[argmax(y)], y[argmax(y)], str(x[argmax(y)]))

grid()
xlabel('Z [mm]')
ylabel('Focus')
title('Best focus @'+ str(x[argmax(y)]))
show()

First install moviepy:

pip install moviepy

Then run this script:

# screech.py
# spectrogram analysis of the screeching motor
# must install moviepy: pip install moviepy
from pylab import *
import moviepy.editor as mp
clip = mp.VideoFileClip("screeching gratings motor.mp4")
d = clip.audio.to_soundarray()
specgram(d[:,0], NFFT=1024, Fs=clip.audio.fps, noverlap=900)
show()

Documentation: https://docs.scipy.org/doc/scipy-1.2.1/reference/

Matrix operations: dot(), inner(), outer(), matrix_power(), det(), trace(), inv(), matrix_rand()

Vector operations: vdot(), norm()

Eigenvalues/eigenvectors: eig()

Solve: solve(), lstsq()

Decompositions: svd(), qr()

Bessel, Airy, etc.

Gaussian quadrature – quad(), Trapezoidal method – trapz(), Simpson – simps(), etc.

from scipy.integrate import quad, simps, trapz
from pylab import *

dh = 1e-3
x = arange(-1, 1, dh)
y = sqrt(1-x**2)

# simple sum
pi1 = 2*sum(y)*dh
print "Using sum, error is ", abs(pi1-pi)

# trapz
pi2 = 2*trapz(y, x)
print "Using trapz, error is ", abs(pi2-pi)

# Simpson
pi3 = 2*simps(y, x)
print "Using simps, error is ", abs(pi3-pi)

# Gaussian quadrature
def circ(x):
    return sqrt(1-x**2)
    
pi4 = 2*quad(circ, -1, 1)[0]
print "Using quad, error is ", abs(pi4-pi)

Including 2-d fft, convolutions, DCT (Discrete Cosine Transform) etc.

Including filter design (IIR, FIR), Wiener filters, convolution/deconvolution, spectral analysis (spectrogram) etc.

from pylab import *
import scipy.signal as signal
t = arange(512)/256.0
y = 5*sin(2*pi*t*0.7)
beats = zeros(size(y))
for i in range(5):
    beats[i*90:i*90+10] = signal.triang(10)
y += beats

subplot(2, 1, 1)
plot(t, y); grid();
ylabel('Signal')

[b, a] = signal.butter(3, 0.05, 'high')
yf = signal.lfilter(b, a, y)
subplot(2, 1, 2)
plot(t, yf); grid()
xlabel('Time[sec]'); ylabel('Signal')
show()

Finding the maximum/minimum.

Exercise:

Exercise: solve the following set of equations using Linear Programming (scipy.optimize.linprog), aka as the Simplex algorithm.

Maximize x+y
Subject to:

1. 4x+3y<=12

2. 3x+4x<=12

3. x=0

4. y>=0

Tip: change “maximize x+y” to minimize “-x-y”

In [3]: c=[-1, -1]
In [4]: A=[[4, 3], [3, 4]]
In [5]: b=[12, 12]
In [6]: so.linprog(c, A, b)
Out[6]:
     fun: -3.4285714285714284
 message: 'Optimization terminated successfully.'
     nit: 2
   slack: array([0., 0.])
  status: 0
 success: True
       x: array([1.71428571, 1.71428571])

Especially useful to read MAT files. The results are stored in a dictionary.

loadmat(), savemat(), whosmat()
Out[57]:
{'__globals__': [],
 '__header__': 'MATLAB 5.0 MAT-file, Platform: PCWIN64, Created on: Sun Mar 03 11:04:50 2019',
 '__version__': '1.0',
 'a': array([[ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10]], dtype=uint8),
 'b': array([u'Nova'], dtype='<U4')}

In [58]: d['a']
Out[58]: array([[ 1,  2,  3,  4,  5,  6,  7,  8,  9, 10]], dtype=uint8)

In [59]: d['b']
Out[59]: array([u'Nova'], dtype='<U4')

pip install wxpython

import wx

class MinimalGUI(wx.Frame):
    """docstring for MinimalFrame"""
    def __init__(self, parent):
        wx.Frame.__init__(self, parent, title="Minimal GUI", size=(640, 480))

app = wx.App()
frame = MinimalGUI(None).Show()
app.MainLoop()

Now let’s add a button. First we must learn about sizers.  See here: https://docs.wxwidgets.org/3.0/overview_sizer.html

import wx
from pylab import *

class MinimalGUI(wx.Frame):
    """docstring for MinimalFrame"""
    def __init__(self, parent):
        wx.Frame.__init__(self, parent, title="Minimal GUI", size=(300, 200))

        # several buttons
        buttonsTxt = ['Plot', 'Talk', 'Exit']
        
        # a sizer object
        sizerH = wx.BoxSizer(wx.HORIZONTAL)
        
        # a list holding all buttons
        self.buttons = []
        for btn in buttonsTxt:
            # we store the button object in the class
            self.buttons.append(wx.Button(self, label=btn))

            # and add it to the panel
            sizerH.Add(self.buttons[-1], border=10, flag=wx.ALL)

            # next, we connect it to a callback function
            self.Bind(wx.EVT_BUTTON, self.OnClick,self.buttons[-1])
        
        # add a text edit for fun
        self.txtCtrl = wx.TextCtrl(self, value="Everyday Python")

        # sizer stuff
        sizerV = wx.BoxSizer(wx.VERTICAL)
        sizerV.Add(sizerH)
        sizerV.Add(self.txtCtrl, border=10, flag=wx.ALL|wx.ALIGN_CENTER)
        sizerV.Fit(self)
        self.SetSizer(sizerV)
        
    # the callback function to the text
    def	OnClick(self, event):
        btnText = event.GetEventObject().GetLabel()
        if(btnText=='Exit'):
            exit()
        if(btnText=='Plot'):
            figure()
            plot(arange(10), sin(arange(10)))
            show()
        if(btnText=='Talk'):
            print "I'm not doing that, ", self.txtCtrl.Value
        
app = wx.App()
frame = MinimalGUI(None).Show()
app.MainLoop()

pip install scikit