Neural network classifiers have problems of over learning

less learning

fall into curse of dimensionality or local minimum

and support vector machine classifiers also have problems of more complex operations

model selection and construction of kernel function is more difficult

and Bayesian classification only in the number of training samples tends to infinity

the training results of the model tends to true.This paper presents a vehicle classification algorithm based on Adaboost.M1.The algorithm is simple to use

and just need to find a weak classifier which′s precision slightly higher than the random prediction

without adjusting any parameters

no prior knowledge

and there is sufficient theoretical support.Finally

experimental results demonstrate the effectiveness of the vehicle classification algorithm based on Adaboost.M1.