search

johnolafenwa / DeepStack

The World's Leading Cross Platform AI Engine for Edge Devices
Apache License 2.0
678 stars 107 forks source link

DeepStack Beta Testing #24

Closed johnolafenwa closed 3 years ago

johnolafenwa commented 4 years ago

In light of recent bugs and issues with no-avx machines. We just pushed two new images for DeepStack

deepquestai/deepstack:gpu-x3-beta and deepquestai/deepstack:cpu-x3-beta

This images should be compatible with no-avx machines as well.

@robmarkcole kindly test this and let's use this thread to address issues ahead on this. Will share this on the forum as well.

robmarkcole commented 4 years ago

Hi John, pulled the cpu version and all fine so far, also shared it on the HA forums. Initial feedback is people are commenting on the slower processing times, would be a good use for the benchmarking script.

Also its surprising it misses the most obvious person here:

image
johnolafenwa commented 4 years ago

Hi @robmarkcole , thanks a lot for testing this and getting feedback too. The speed on cpu is currently not optimised, the gpu speed should be about the same though. The final version is going to be fast enough. Performance optimisation will be done before the stable release.

I am quite surprised about the image, you can test with a value of about 0.65 for the min_confidence parameter.

Would adjust the default confidence as well. Kindly share the original image.

robmarkcole commented 4 years ago

Processing 15 images with noavx took 60.3 seconds with total of 91 predictions. With cpu-x3-beta took 137.8 seconds with a total of 104 predictions.

The pool image is here.

johnolafenwa commented 4 years ago

Thanks @robmarkcole , we should be able to get the new version to run as fast as the current version. Will be making a couple of optimizations soon.

robmarkcole commented 3 years ago

stale