Skip to main content
. Author manuscript; available in PMC: 2014 Jul 1.
Published in final edited form as: Int J Comput Vis. 2013 Feb 19;103(3):348–371. doi: 10.1007/s11263-013-0609-0

Fig. 13. Feature selection with aLDA.

Fig. 13

Each plot shows the per-category recognition rate for different combinations of features. The training rate is shown on the left with a darker bar. The test set is shown on the right with a lighter bar. The two numbers displayed right after the feature combination label are the training and test recognition rates averaged across material categories. Our feature selection algorithm finds “color + SIFT + edge-slice” to be the optimal feature set for the aLDA model on FMD images.