Model Zoo

InsightFace Model Zoo

1. InsightFace/ArcFace recognition models.

LResNet100E-IR|Emore@BaiduDrive, Trained on insightv2/Emore dataset. Performance:

Method	LFW(%)	CFP-FF(%)	CFP-FP(%)	AgeDB-30(%)	MegaFace(%)
Ours	99.77	-----	98.27	98.28	98.47

MobileFaceNet@BaiduDrive Performance:

Method	LFW(%)	CFP-FF(%)	CFP-FP(%)	AgeDB-30(%)	MegaFace(%)
Ours	99.50	-----	88.94	95.91	-----

LResNet50E-IR@BaiduDrive, @GoogleDrive Performance:

Method	LFW(%)	CFP-FF(%)	CFP-FP(%)	AgeDB-30(%)	MegaFace(%)
Ours	99.80	99.83	92.74	97.76	97.64

LResNet34E-IR@BaiduDrive Performance:

Method	LFW(%)	CFP-FF(%)	CFP-FP(%)	AgeDB-30(%)	MegaFace(%)
Ours	99.65	99.77	92.12	97.70	96.70

Caffe LResNet50E-IR@BaiduDrive, converted by above MXNet model. Performance:

Method	LFW(%)	CFP-FF(%)	CFP-FP(%)	AgeDB-30(%)	MegaFace1M(%)
Ours	99.74	-TBD-	-TBD-	-TBD-	-TBD-

2. Gender&Age models.

ResNet50x2@BaiduDrive, trained on Asian dataset.

3. Face detection models.

SSH modelv1 on BaiduDrive and GoogleDrive

Evaluation on WIDERFACE:

Impelmentation	Easy-Set	Medium-Set	Hard-Set	Speed
MTCNN	0.851	0.820	0.607	CPU realtime
Original Caffe SSH	0.93123	0.92106	0.84582	GPU realtime
Our SSH Modelv1	0.93394	0.92187	0.83682	GPU realtime