Random Forest with Adaptive Local Template for Pedestrian Detection

Algorithm 2

Procedure of assembling weak classifiers.

Input:

 Number of trees:

 Maximum depth:

 Training set for each tree:

 Template set for each tree:

(1) Initialize weights for samples of each tree:

(2) Compute class distribution of root node of each tree by (6)

(3) for   to   do

(4) Check stopping criterions for nodes in depth

(5) Split nodes in depth by Algorithm

(6) Update weight by (13) for each sample in each

(7) Calculate class distribution of newly created nodes in depth by (6)

(8) end for