Item — HN Reader

Inference times for inception-v3 on a Raspberry Pi have been benchmarked recently [1]. They weigh in at around 2s. If you remove the python layer it gets down to 500ms. However, it's not yet clear if the python overhead is the only reason for that.

Nvidia's Tegra X1 should supposedly be capable of <10ms for imagenet grade models [2]. It's fair to assume though, that this must be for trimmed down and/or 16bit models as compared to full inception models.

And finally, Sam who also facilitated building TF on the Pi is about to host a 6 weeks half theory, half practice course on TF and deep learning [3] (me thinks he deserves this plug).

[1] https://github.com/samjabrahams/tensorflow-on-raspberry-pi/t...

[2] https://youtu.be/_4tzlXPQWb8?t=53m35s

[3] https://www.thisismetis.com/deep-learning-with-tensorflow

4 comments

Does this use the Pi's GPU? It's not clear and I think the answer is no?

That's really unlikely. There's no OpenCL for the RPi. GPGPU programming for the Pi is mostly assembler based:

https://github.com/nineties/py-videocore

https://petewarden.com/2014/08/07/how-to-optimize-raspberry-... (example involves deep learning!)

https://www.raspberrypi.org/forums/viewtopic.php?f=29&t=7891... (someone actually tried making an LLVM backend!)

I really doubt someone integrated that into TensorFlow…

We only used the Pi's CPU, which is adequate for classifying images, but not powerful enough to train a model

Hey Alex, thanks for the comment! Have you tried out image classification with the Tegra X1 yourself by chance??

I haven't but the ressources on jetsonhacks.com are amazing to see robot vision on this board in action.

If we get down to <10ms does that mean that real-time video processing becomes feasible?

But real life images are at least 640*480, not imagenet.

The training/testing data for ImageNet are not all tiny thumbnails. It's just that the models work better/are more cleanly defined with uniform inputs, so images are scaled/cropped down to the 224x224 or 299x299 or whatever dimensions. Inception-ResNet works fine on huge images, it just doesn't use all of the pixel data.