Re:TensorFlow Image Recognition on a Raspberry Pi
by pjc50 on Thursday February 09, 2017 @07:41AM
(#13606490)
Does this use the Pi's GPU? It's not clear and I think the answer is no?
Re:TensorFlow Image Recognition on a Raspberry Pi
by floatboth on Thursday February 09, 2017 @10:58AM
(#13607792)
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…
Re:TensorFlow Image Recognition on a Raspberry Pi
by mrubashkin on Thursday February 09, 2017 @02:53PM
(#13610290)
We only used the Pi's CPU, which is adequate for classifying images, but not powerful enough to train a model
Re:TensorFlow Image Recognition on a Raspberry Pi
by mrubashkin on Thursday February 09, 2017 @02:55PM
(#13610313)
Hey Alex, thanks for the comment! Have you tried out image classification with the Tegra X1 yourself by chance??
Re:TensorFlow Image Recognition on a Raspberry Pi
by alex_hirner on Thursday February 09, 2017 @08:11PM
(#13612528)
I haven't but the ressources on jetsonhacks.com are amazing to see robot vision on this board in action.
Re:TensorFlow Image Recognition on a Raspberry Pi
by iaw on Thursday February 09, 2017 @12:16AM
(#13604842)
If we get down to <10ms does that mean that real-time video processing becomes feasible?
Re:TensorFlow Image Recognition on a Raspberry Pi
by deepnotderp on Thursday February 09, 2017 @12:58PM
(#13609128)
But real life images are at least 640*480, not imagenet.
Re:TensorFlow Image Recognition on a Raspberry Pi
by sja on Thursday February 09, 2017 @01:53PM
(#13609686)
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.
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