Table 2. The classification results of Facenet model under different processing methods.

Method	Recall (%)	Precision (%)	Accuracy (%)	F1 (%)
Original	70.6	65.4	75.3	67.9
Pixelated	30.1	29.3	33.6	29.7
Masked	5.64	6.89	6.17	6.2
CIAGAN	26.3	24.8	29.4	25.5
Ours	40.8	38.9	43.7	39.8