Data collection for both training and testing a classifier is a tedious but essential step towards face detection and recognition. All of the statistical methods suffer from this problem. This paper presents a genetic algorithm (GA)-based method to swell face database through re-sampling from existing faces. The basic idea is that a face is composed of a limited components set, and the GA can simulate the procedure of heredity. This simulation can also cover the variations of faces in different lighting conditions, poses, accessories, and quality conditions. All the collected face samples are aligned and randomly divided into three sub-sets: training, validating, and testing set. The training set is then used to train a sparse network of winnow (SNoW). In addition, it is also used as the initial population of the GA. After each generation, we use the initial generation and the solutions with high fitness values to re-train the SNoW, and the newly-trained SNoW is used to evaluate the individuals of next generation and also tested on validation set and test set. To verify the generalization capability of the proposed method, we also use the expanded database to train an AdaBoost-based face detector and test it on the MIT+CMU frontal face test set. The experimental results show that the data collection can be speeded up efficiently by the proposed methods.