search

shamangary / FSA-Net

[CVPR19] FSA-Net: Learning Fine-Grained Structure Aggregation for Head Pose Estimation from a Single Image
Apache License 2.0
612 stars 155 forks source link

[Training Time] #32

Closed leviethung2103 closed 5 years ago

leviethung2103 commented 5 years ago

Hell Authors,

Thank you for such a great project.

I would like to know the training time if I want to train from the scratch and produce the same results like you. Furthermore, how many GPU 1080ti you used for research ?

Thank you so much

shamangary commented 5 years ago

The training can be done within 3 hrs. 1 gpu 1080ti is more than enough for this project. If you choose smaller batchsize it will cost more.