Install and setup anaconda python

Download Anaconda and run the installer. This can be done using the graphical installer on a local laptop/workstation or if you are logged into a linux computing server use:

# change the following url to the most recent one from
wget #download the installer
bash #run the installer script
export PATH=/path/to/anaconda3/bin:$PATH #change this pathname accordingly and append to your shell path so you can use conda to create environment below

Then from your command-line environment (e.g. linux Terminal on OS X, check out the unix tricks/tutorial), create a new python environment:

conda create --name py35 python=3 anaconda
#conda create --name py27 python=2.7 anaconda #if you want to install python 2.7 use this line instead. useful for opencv usage on linux cluster

Install extra modules with conda

Next, activate your new environment and install a couple add on modules:

source activate py35
conda install opencv
conda install h5py
conda install tifffile -c conda-forge
conda install joblib
computer vision library
read/write support for hdf5
read matlab v7.3 files
fixes other hdf5 dependencies (like cv2)
read in multi-page tiff files (time-lapse movies or multi-channel images)
though is already a part of scikit-image, so this extra installation may not be necessary.
a parallel computing module
useful with multicore compute nodes

Run jupyter notebook/ipython

Then whenever you want to use your python 3 environment, first activate your environment (if it's not already active-- source activate py35) and enter:

jupyter notebook

Then from the webserver jupyter notebook/ipython session in your web browser click on new Python notebook:

This will open a new python session where you paste and run code (control-enter evaluates the current cell, shift-enter evaluates and moves to a new cell):

import cv2
import numpy as np

print("OpenCV Version : {0}".format(cv2.__version__))
print("Numpy Version : {0}".format(np.__version__))

OpenCV Version : 3.1.0
Numpy Version : 1.10.4

To shut down your session go to File --> Close and Halt.

Then from your command-line shell from which the webserver is running, press control-c to interrupt/stop.