000_index.ipynb

Kernel: Python 3

Image analysis in Python with SciPy and scikit-image

Please clone

https://github.com/scikit-image/skimage-tutorials

scikit-image is a collection of image processing algorithms for the SciPy ecosystem. It aims to have a Pythonic API (read: does what you'd expect), is well documented, and provides researchers and practitioners with well-tested, fundamental building blocks for rapidly constructing sophisticated image processing pipelines.

In this tutorial, we provide an interactive overview of the library, where participants have the opportunity to try their hand at various image processing challenges.

The tutorial consists of four parts, in which we:

1. give a general overview of the functionality available in the various submodules;

2. showcase analysis of real-world, anisotropic three-dimensional microscopy data, in celebration of the 0.13 release which greatly improves N-dimensional processing;

3. demonstrate how to use scikit-image for machine learning tasks in combination with scikit-learn, and

4. highlight interaction with other libraries, such as Keras and SciPy's LowLevelCallable.

Attendees are expected to have a working knowledge of NumPy, SciPy, and Matplotlib.

Across domains, modalities, and scales of exploration, images form an integral subset of scientific measurements. Despite a deep appeal to human intuition, gaining understanding of image content remains challenging, and often relies on heuristics (we hope to address some of that in part 3!). Even so, the wealth of knowledge contained inside of images cannot be understated, and scikit-image, along with SciPy, provides a strong foundation upon which to build algorithms and applications for exploring this domain.

Prerequisites

All of the below packages, including the non-Python ones, can be found in the Anaconda Python distribution, which can be obtained for free. (Though some packages may need to be installed using conda install packagename.)

Required packages

• scikit-image (0.13)

Required for scikit-image:

• Python >= 3.5

• numpy >= 1.13.1

• scipy >= 0.19

Required for image viewing and other examples:

• matplotlib >= 2

Required for machine learning section:

• scikit-learn >= 0.18

Example images

scikit-image ships with some example images in skimage.data.

Sections

For convenience, we have divided this tutorial into several chapters, linked below. Throughout the tutorials, feel free to ask questions. We want you to come away confident in your image analysis skills!

• 1:30 - 1:40: Introduction

• 1:40 - 2:10: Images are just NumPy arrays, RANSAC

• 2:15 - 3:30: 3D Image Processing

• 3:35 - 4:50: Machine Learning

• 4:55 - 5:30: Other Libraries

Note: Snacks are available 2:15-4:00; coffee & tea until 5.

For later

• Check out the other lectures

• Try some StackOverflow challenges

Further information

• Here are some real world use cases.

• The other scikits & interoperation with other Python packages

• ndimage vs scikit-image

• scikit-image vs opencv

After the tutorial

Stay in touch!

• Follow the project's progress on GitHub.

• Ask the team questions on the mailing list

• Contribute!

• Read our paper (or this one, for skimage in microscopy)

import numpy as np
np.__version__

'1.13.0'

!pip install numpy --upgrade

Collecting numpy
  Downloading numpy-1.13.1-cp36-cp36m-manylinux1_x86_64.whl (17.0MB)
    100% |████████████████████████████████| 17.0MB 90kB/s eta 0:00:01    63% |████████████████████▏           | 10.7MB 1.7MB/s eta 0:00:04    83% |██████████████████████████▋     | 14.1MB 898kB/s eta 0:00:04
Installing collected packages: numpy
  Found existing installation: numpy 1.13.0
    Uninstalling numpy-1.13.0:
      Successfully uninstalled numpy-1.13.0
Successfully installed numpy-1.13.1

import skimage

skimage.__version__

'0.13.0'