4 ms·

Does anyone know an alternative that works on RaspberryPi? This states: "Detectron operators currently do not have CPU implementation; a GPU system is required.

Does anyone know an alternative that works on RaspberryPi? This states: "Detectron operators currently do not have CPU implementation; a GPU system is required." Even low FPS (3-5) would be acceptable.

m_ke 9y ago
You could try tensorflow object_detection api with tensorflow lite https://github.com/tensorflow/models/tree/master/research/object_detection https://github.com/tensorflow/models/tree/master/research/ob... google also recently put up their mobilenet v2 paper which handles segmentation https://arxiv.org/abs/1801.04381 https://arxiv.org/abs/1801.04381
- deepGem 9y ago
  +1 for the Google object detection API. The trained model is quite huge though. 200 MB based on Resnet faster R-CNN. There are creative ways of chunking this model to keep it small.
  - m_ke 9y ago
    I think they have a mobilenet SSD model as well.
pilooch 9y ago
See https://github.com/jolibrain/dd_performances https://github.com/jolibrain/dd_performances you can reach 1fps on image classification and objects detection tasks on rpie3 with custom caffe and models.
haeffin 9y ago
It's a bit disappointing, when caffe2 was released it was stated that mobile is a big focus, but things like this don't support mobile (even though some of this was demoed by FB on a phone).
- ajtulloch 9y ago
  FWIW we have some similar R-CNN models using this codebase running in the FB apps on iOS/Android with Caffe2 - with some neat bells and whistles (eg full execution of the compute graph on the device GPU where available via Metal/OpenGL). I’ll look into adding a mobile tutorial for Detectron if you’re interested.
  - lovelearning 9y ago
    Not the OP, but +1 from me for such a mobile tutorial. And thank you in advance!
  - haeffin 9y ago
    That would be awesome. But - is a tutorial enough? From what I gather from the readme the custom ops have cuda implementations and no CPU ones, which you'd like to have as a fallback on mobile (even if metal/opengl implementations exist), or am I wrong?
    - ajtulloch 9y ago
      They exist and are (nearly all) open-sourced for CPU & Metal (I wrote some of them).
      - haeffin 9y ago
        Nice - the "Detectron operators currently do not have CPU implementation; a GPU system is required." on https://github.com/facebookresearch/Detectron/blob/master/INSTALL.md https://github.com/facebookresearch/Detectron/blob/master/IN... is outdated then? Or is it the "nearly all" that's behind this statement?
        stzpz 9y ago
        We have open sourced several operators for detectron so that the model could run on GPU/CPU/mobile (https://github.com/facebookresearch/Detectron/commit/757d77c239fbef9d6a341cdfcea2f43219175709 https://github.com/facebookresearch/Detectron/commit/757d77c...). We also open sourced a conversion tool (https://github.com/facebookresearch/Detectron/commit/757d77c239fbef9d6a341cdfcea2f43219175709 https://github.com/facebookresearch/Detectron/commit/757d77c...) to help converting the model.
vitorgrs 9y ago
There's CNTK, but I don't know how well does it work :) https://github.com/Microsoft/CNTK https://github.com/Microsoft/CNTK
fmntf 9y ago
With such deep networks, I think it would be hard to get 3-5 FPS on an Intel iX; forget the Raspberry PI CPU!
byte1918 9y ago
Maybe https://pjreddie.com/darknet/install https://pjreddie.com/darknet/install or one of the various forks. YOLO is supposed to be one of the best object detection systems as far as accuracy/speed is concerned.
- ajtulloch 9y ago
  Look at the first table in https://arxiv.org/pdf/1708.02002.pdf https://arxiv.org/pdf/1708.02002.pdf (the RetinaNet paper, code + models included in this release). There’s a substantial accuracy cost for YOLO vs other, similarly fast (within a factor of 2 or so), methods.
  - byte1918 9y ago
    > similarly fast Unless I misunderstood figure 2, YOLO seems to be more than twice as fast than the second runner and yes I agree and already was aware that YOLO's accuracy is not as good as other classifiers.
kurtisc 9y ago
FWIW, A Pi does have a GPU with a full OpenGL ES implementation. However, this requires NVIDIA CUDA.
eggie5 9y ago
i ras running a NAS-based object detection model on my big-new MacBook and it was taking about 30s/image on an unoptimised tensorflow build. I then tried a model/net model which took about 3-4s/image.
dapreja 9y ago
Use your last android phone.

m_ke 9y ago

You could try tensorflow object_detection api with tensorflow lite https://github.com/tensorflow/models/tree/master/research/object_detection https://github.com/tensorflow/models/tree/master/research/ob... google also recently put up their mobilenet v2 paper which handles segmentation https://arxiv.org/abs/1801.04381 https://arxiv.org/abs/1801.04381

deepGem 9y ago
+1 for the Google object detection API. The trained model is quite huge though. 200 MB based on Resnet faster R-CNN. There are creative ways of chunking this model to keep it small.
- m_ke 9y ago
  I think they have a mobilenet SSD model as well.

pilooch 9y ago

See https://github.com/jolibrain/dd_performances https://github.com/jolibrain/dd_performances you can reach 1fps on image classification and objects detection tasks on rpie3 with custom caffe and models.

haeffin 9y ago

It's a bit disappointing, when caffe2 was released it was stated that mobile is a big focus, but things like this don't support mobile (even though some of this was demoed by FB on a phone).

ajtulloch 9y ago
FWIW we have some similar R-CNN models using this codebase running in the FB apps on iOS/Android with Caffe2 - with some neat bells and whistles (eg full execution of the compute graph on the device GPU where available via Metal/OpenGL). I’ll look into adding a mobile tutorial for Detectron if you’re interested.
- lovelearning 9y ago
  Not the OP, but +1 from me for such a mobile tutorial. And thank you in advance!
- haeffin 9y ago
  That would be awesome. But - is a tutorial enough? From what I gather from the readme the custom ops have cuda implementations and no CPU ones, which you'd like to have as a fallback on mobile (even if metal/opengl implementations exist), or am I wrong?
  - ajtulloch 9y ago
    They exist and are (nearly all) open-sourced for CPU & Metal (I wrote some of them).
    - haeffin 9y ago
      Nice - the "Detectron operators currently do not have CPU implementation; a GPU system is required." on https://github.com/facebookresearch/Detectron/blob/master/INSTALL.md https://github.com/facebookresearch/Detectron/blob/master/IN... is outdated then? Or is it the "nearly all" that's behind this statement?
      - stzpz 9y ago
        We have open sourced several operators for detectron so that the model could run on GPU/CPU/mobile (https://github.com/facebookresearch/Detectron/commit/757d77c239fbef9d6a341cdfcea2f43219175709 https://github.com/facebookresearch/Detectron/commit/757d77c...). We also open sourced a conversion tool (https://github.com/facebookresearch/Detectron/commit/757d77c239fbef9d6a341cdfcea2f43219175709 https://github.com/facebookresearch/Detectron/commit/757d77c...) to help converting the model.

vitorgrs 9y ago

There's CNTK, but I don't know how well does it work :) https://github.com/Microsoft/CNTK https://github.com/Microsoft/CNTK

fmntf 9y ago

With such deep networks, I think it would be hard to get 3-5 FPS on an Intel iX; forget the Raspberry PI CPU!

byte1918 9y ago

Maybe https://pjreddie.com/darknet/install https://pjreddie.com/darknet/install or one of the various forks. YOLO is supposed to be one of the best object detection systems as far as accuracy/speed is concerned.

ajtulloch 9y ago
Look at the first table in https://arxiv.org/pdf/1708.02002.pdf https://arxiv.org/pdf/1708.02002.pdf (the RetinaNet paper, code + models included in this release). There’s a substantial accuracy cost for YOLO vs other, similarly fast (within a factor of 2 or so), methods.
- byte1918 9y ago
  > similarly fast Unless I misunderstood figure 2, YOLO seems to be more than twice as fast than the second runner and yes I agree and already was aware that YOLO's accuracy is not as good as other classifiers.

kurtisc 9y ago

FWIW, A Pi does have a GPU with a full OpenGL ES implementation. However, this requires NVIDIA CUDA.

eggie5 9y ago

i ras running a NAS-based object detection model on my big-new MacBook and it was taking about 30s/image on an unoptimised tensorflow build. I then tried a model/net model which took about 3-4s/image.

dapreja 9y ago

Use your last android phone.