Geek Answers Handbook

Using Weka in Images

I am new to Weka, and from examples of how to use it, I only saw text problems. Can I use images in Weka with machine learning classifiers?

+4
source share
3 answers

You can directly classify pixels using the Trainable Weka Segmentation plugin (formerly the advanced Weka Segmentation plugin) from Fiji/ImageJ .

The plugin is designed for segmentation through interactive learning . This means that the user must select a set of functions (edge ​​detectors, texture filters, etc.), select the number of classes (2 by default) and interactively draw (with ROI tools) samples of all classes. After training the classifier based on these samples, all the pixels in the image will be classified, and the segmentation result will be displayed by overlaying the original image. The idea is to repeat this process (drawing + training) until satisfactory segmentation is obtained.

The plugin also provides a set of tools for saving / loading samples in the ARFF format and saving / loading the classifier in .model format, so it is fully compatible with the latest version of WEKA.

If you want to do image classification , you can also reuse some of the plugin methods.

+7
source

You can use an open source image processing application such as ImageJ and Fiji to extract functions from your image and use it in Weka

Fiji has a plugin called Advanced Weka Segmentation , which should be very useful when applying Weka classifiers to Image

+3
source

Weka machine learning classifiers work with numerical and categorical features. Before using weka with images, you need to extract functions from your images. There are quite simple functions, such as average, maximum, average, to suit your needs. Or you may need to use some other algorithms for your images.

The following are Wikipedia feature extraction algorithms.

Low level

  • Edge detection
  • Corner detection
  • blob detection
  • Ridge detection
  • Scale Invariant Function Conversion

I suggest reading an overview of optical character recognition to understand how they are used. OCR is a fairly simple use case. OCR has standard data sets and algorithms. Therefore, it is very instructive to learn about it.

+2
source

More articles:


All Articles