Reinforced AdaBoost Learning for Object Detection with Local Pattern Representations

Figure 4

Selection of weak classifier for each round in AdaBoost learning. At first, LPR images are extracted from the training images. The histogram of LPR is generated by accumulative weights of samples according to the LPR feature values of the same pixel location. Assuming that the 1st pixel location is the best pixel location with the smallest error rate, it can be selected as a weak classifier for that round in AdaBoost learning. The selected weak classifier is described as a lookup table which is generated by (8), and the real values in the lookup table represent the confidence level of prediction.