Result of algorithm Multimodal S-MAD 1 on SMAD-BIOLAB-1.0

Accuracy indicators
EER	BSCER10	BSCER20	BSCER100
40.45% (40.00% - 40.91%)	0%	0%	0%

Detection failures
REJNBFRA	REJNMRA
0%	0%

Efficiency indicators
Avg Detection Time	Avg Detection Time (BF)	Avg Detection Time (M)
551.6 s	551.6 s	551.6 s

Memory indicators

Max Detection Memory

Not measured

Graphs
Score distributions	MACER(t) and BSCER(t) graphs	DET graph
Score distributions	MACER(t) and BSCER(t) graphs	DET graph

Description of algorithm Multimodal S-MAD 1

Algorithm type: Win32 executable

A - Brief description

Algorithm is based on the saliency regions (full face, eyes, nose and mouth) extracted from the face image. Each of the multimodal regions are processed independently using the multi-level fusion of multiple feature extraction and classification techniques.

B - References

R. Ramachandra et al., "Multimodality for Reliable Single Image based Face Morphing Attack Detection", submitted to IEEE ACCEESS, 2022.