Performence issue(?) when running 2 requests at the same time


#21

Hello @elad.bar Sorry about the late reply. We are working on a major update with support for batch requests and more effective batching for single requests, enabling you to effectively control the cpu and gpu utilization.
This will be released very soon, apologies for the late update.


#22

thanks for the update,
I saw that new version uploaded to docker hub, it seems that now it takes ~480 msec for single request every 2 seconds as oppsed to ~290 msec before.


#23

Hello @elad.bar, the update fixed a critical bug, on the latency increase, we have made some improvements in the face api, in the next few days, we are releasing an update that will provide magnitude improvements in speed for the face apis. You will love the speed.


#24

Hi @elad.bar , we have released a new update to the gpu version. The face detection api is now a lot faster and we have improved the speed for the recognition as well.