deepinsight / deepinsight/insightface
Accuracy on cfp_fp is too low.
- Dominant language
- Python
- Stars
- 29.7k
- Forks
- 6.1k
- PR merge metrics
- No merged PRs in 30d
Description
I am training vargfacenet with inital learning rate as 0.01 because learining rate 0.1 gives nan loss.
Other accuracy seems fine but cfp_fp accuracy is too low.
dataset : retina
testing verification..
(12000, 512)
infer time 10.723083
[lfw][84000]XNorm: 11.444277
[lfw][84000]Accuracy-Flip: 0.99117+-0.00460
FAR : 0.0013333333333333335
testing verification..
(14000, 512)
infer time 11.187130000000005
[cfp_fp][84000]XNorm: 270.588836
[cfp_fp][84000]Accuracy-Flip: 0.68929+-0.01869
FAR : 0.001142857142857143
testing verification..
(12000, 512)
infer time 9.597576
[agedb_30][84000]XNorm: 10.858622
[agedb_30][84000]Accuracy-Flip: 0.91533+-0.01802
FAR : 0.0013333333333333335
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.