DeepFakes Detection Based on Heart Rate Estimation: Single- and Multi-frame

Abstract This chapter describes a DeepFake detection framework based on physiological measurement. In particular, we consider information related to the heart rate using remote photoplethysmography (rPPG). rPPG methods analyze video sequences looking for subtle color changes in human skin, revealing presence of blood under tissues. explores what extent is useful videos. We recent fake detector named DeepFakesON-Phys that Convolutional Attention Network (CAN), which extracts spatial and temporal from frames, analyzing combining both sources better detect has been experimentally evaluated latest public databases field: Celeb-DF v2 DFDC. The results achieved single frame are over 98% AUC (Area Under Curve) databases, proving success detectors measurement this chapter, also propose study heuristical statistical approaches performing continuous by scores consecutive frames with low latency high accuracy (100% evaluation dataset). show extracted short-time can improve discrimination power DeepFakesON-Phys.