Table 4: Performance of detection on MIT + CMU test set of ABANN detector.

NameAdaBoost structure
Number of strong classifiersNumber of Haar-like featuresANN structureFace detectedMissed faces (false rejection)False detectionsDetection rateAverage time to process an image (second)

AB-ANN201925Rowley’s model466411391.91%0.174
AB-ANN25291345057388.76%0.195