Skip to content

ESE 650: Learning in Robotics Project 1, Color Segmentation using Gaussian Mixture Models

License

Notifications You must be signed in to change notification settings

NitinJSanket/ESE650Project1

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Colored Barrel Detection Using Gaussian Mixture Model based Color Segmentation

Problem Statement

Given a set of training images in the folder Train with the file names as the distance to the barrel in meters, locate all the barrels and find the distance to them from the camera in each Test image.

Usage Guide:

  1. Run Wrapper.m
  2. Change ColorSpace by changing ColorSpace variable to any one of RGB or YCbCr or yRb. The code thresholds currently work for YCbCr.
  3. To train, Set TrainFlag = 1, switched off by default.
  4. If you have any questions or queries feel free to raise an issue.

Report:

You can find the report here.

Sample Input and Outputs:

Yellow highlights show the candidate red regions and green highlights show the candidate red barrel regions.

GMM Covariance while being trained:

GMM Training Video

Reference Codes:

  1. Oriented Bounding Box http://www.mathworks.com/matlabcentral/fileexchange/30402-feret-diameter-and-oriented-box/content/imFeretDiameters/imFeretDiameter.m
  2. Ellipsoid Plotting http://www.mathworks.com/matlabcentral/fileexchange/13844-plot-an-ellipse-in--center-form-
  3. Entropy rate segmentation http://www.merl.com/publications/docs/TR2011-035.pdf

Releases

No releases published

Packages

No packages published