Face detection from images is a key problem in human computer interaction studies and pattern recognition research. In this work, we propose an efficient genetic algorithm (EGA) that solves the face detection problem in color images. The proposed EGA is based on the Takagi-Sugeno-Kang(TSK)-type fuzzy model employed to perform parameter learning. Compared with traditional genetic algorithms, the EGA uses the sequential-search based-efficient generation (SSEG) method to generate an initial population to determine the most efficient mutation points. Experimental results show that the performance of the EGA is superior to the existing traditional genetic methods.