deepinsight / deepinsight/insightface
Training on Retina-Face aligned database
Open
- Dominant language
- Python
- Stars
- 29.7k
- Forks
- 6.1k
- PR merge metrics
- No merged PRs in 30d
Description
Hi,
I have questions concerning the connection between the alignment pipeline and its influence on the recognition performance.
1. Which pipeline are you using for the FRVT submission? Training on MTCNN aligned images with Resnet 101/Arcface loss and using Retina Face only in the inference step?
2. It seems that you only provide MTCNN-aligned training DBs, wouldn´t it make sense to use Retina Face alignment also for the training DBs (not only at the test/production stage) ?
3. Are there benchmarks showing the influence of the alignment pipeline on the overall recognition performance?
Kind regards,
Christian
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.