Random Forest with Adaptive Local Template for Pedestrian Detection

Algorithm 1

Learning splitting function with local template selector.

Input:

 Samples at a node in :

 Exemplar set of :

 Block size of each sample:

 Maximum size of local template:

Output:

 Optimal splitting parameter:

(1) Initialization:

(2) for  each   do

(3) for   to   do

(4)  Randomly generate configuration:

(5)  Generate adaptive local DOT template in according to

(6)  Calculate the maximum and minimum value of :

(7)  for   to   do

(8)   Randomly select a threshold

(9)    Divide samples at into two subset:

(10)   Calculate class-label purity by (3)

(11)     if  

(12)     

(13)     , , , , ,

(14)     end if

(15)  end for

(16) end for

(17) end for