Adaboost人脸检测方法的改进

针对Adaboost人脸检测训练非常耗时的问题，从训练中直接求解目标函数和弱分类器使用双阈值判决构造强分类器两个方面对人脸检测系统进行了改进。实验结果表明，改进后的系统使用的弱分类器数目大大减少，并且训练速度比传统方法高11倍左右。

Improvement of Adaboost face detection

Aiming at the problem that training time of Adaboost face detection is extremely long, two improvement methods were proposed:One method was to directly solve the parameter of single weaker classifier,the other was to introduce a double threshold decision to make stronger classifier.The experiment results show that the number of weaker classifiers needed in Adaboost face detection system updated is dramatically reduced and its training speed is about 11 times higher than that of the traditional method.