Neural Network ModelsΒΆ

Here you can see a summary for neural network models created by our Lab and used in FindFace Enterprise Server SDK:

Note

The CPU and GPU benchmark setup is the following:

  • CPU: OpenBLAS 0.2.18 (single thread), Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz
  • GPU: CUDA 8.0, GeForce GTX 1080
Type Name In use Facen size CPU GPU Accuracy (Black-dirty)
FPS Memory, MB FPS Memory, MB
Face biometrics model_36 2016 160 N/A N/A N/A N/A N/A
model_39c 2016 160 N/A N/A N/A N/A N/A
fr_1 2016-04/05/2017 160 N/A N/A N/A N/A N/A
en_1 2016-03/03/2017 320 N/A N/A N/A N/A N/A
en2_face0 since 03/14/2017 320 N/A N/A N/A N/A N/A
apricot_160f since 07/31/2017 160 5.15 336 62.27 622 54.48
apricot_320 since 07/31/2017 320 4.87 386 59.92 616 59.32
banana_800f since 09/15/2017 800 0.71 2407 11.96 2638 61.25
cherry_480 since 03/30/2018 480 1.24 1880 26.45 2144 71.91
Gender recognition fr_1_gender0 since 04/05/2017 N/A N/A N/A N/A N/A N/A
Age recognition fr_1_age0 since 04/05/2017 N/A N/A N/A N/A N/A N/A
Emotions recognition model_39c_em 04/05/2017- 08/11/2017 N/A N/A N/A N/A N/A N/A
emotion_1 since 08/11/2017 N/A N/A N/A N/A N/A N/A