Difference between revisions of "Refactoring of the classification chain"

From OTBWiki
Jump to: navigation, search
(Created page with "Currently, the [http://wiki.orfeo-toolbox.org/index.php/Classification_chain classification tools] give very decent performances in most cases with the default setup, which is...")
 
Line 19: Line 19:
 
* current one is balanced random
 
* current one is balanced random
 
=== Allow for the generation of disjoint training and validation sample sets ===
 
=== Allow for the generation of disjoint training and validation sample sets ===
* Define normalisation strategies  
+
=== Define normalisation strategies ===
** centered-reduced
+
* centered-reduced
** min-max
+
* min-max
** other  
+
* other  
 
=== Define stynthetic sample generation strategies ===
 
=== Define stynthetic sample generation strategies ===
 
* add random noise
 
* add random noise
 
* combine exiting samples
 
* combine exiting samples
 
* other
 
* other

Revision as of 08:23, 4 October 2013

Currently, the classification tools give very decent performances in most cases with the default setup, which is nice, but they lack fine tuning and flexibility especially for those samples related steps before calling the training algorithms.

These are some of the issues that cuold be taken into account:

Sample source:

  • images
  • vector data (GIS files)
  • CSV files
  • other

Split sampling, sample normalisation, learning and validation (at the application level)

  • needs the definition of sample I/O format and drivers

Define sample strategies

  • raw vs balanced
  • systematic (1/N) vs random vs stratified
  • current one is balanced random

Allow for the generation of disjoint training and validation sample sets

Define normalisation strategies

  • centered-reduced
  • min-max
  • other

Define stynthetic sample generation strategies

  • add random noise
  • combine exiting samples
  • other