AdaBoost based training method has become a state-of-the-art boosting approach in face detection system. In this paper, compared to the naive AdaBoost method, Forward Feature Selection (FFS) method is used in feature selection to reduce the training time by about 50 to 100 times without loss of performance. Furthermore, hierarchical feature spaces (both local and global) to construct a detector cascade based on FFS method are adopted, which still have good discrimination in the later stage of boosting process. Experimental results show that our method can achieve higher performance using far less training time.
展开▼
