These results are derived by strictly following the FRGC version 2 protocols: (i) use only the 12,776 training images for training; and (ii) generate the performance curves using the Biometric Experimentation Environment (BEE). We apply only one training process for both FRGC Experiment 4 and FRGC Experiment 1, i.e., one method with identical parameters for both experiments. (SPONSORS AND VENTURE CAPITAL INVESTORS ARE WELCOME FOR COMMERCIALIZING OUR FACE DETECTION AND RECOGNITION TECHNOLOGIES)

FRGC Experiment 4 (ROC III, generated by BEE)



FRGC Experiment 1 (ROC III, generated by BEE)