any trained model deployment doc?

[wswfc]

Hi,

I'm just starting getting into caffe and am walking through the tutorials on the site. Here is one question that I wish anyone could shed some lights on:

In the simplest example doc: http://caffe.berkeleyvision.org/gathered/examples/mnist.html

it says:

The final model, stored as a binary protobuf file, is stored at

lenet_iter_10000

which you can deploy as a trained model in your application, if you are training on a real-world application dataset.

But how? how can I use such model for deployment purpose? I couldn't find any doc related to it. Seems the "caffemodel file" way is deprecated now that all we got is a single file?

[Przemek D]

For ways to load a trained model and query it - see the examples.

[wswfc]

Hi Przemek D,

Thanks for your help. So looks like the doc in http://caffe.berkeleyvision.org/gathered/examples/mnist.html is not accurate. The training will generate .caffemodel file and to use it, we need to modify the net prototxt file to a deploy.prototxt file? Is that understanding correct?

在 2017年10月11日星期三 UTC+8下午6:45:39，Przemek D写道：

[Jonathan R. Williford]

It looks like the tutorial for MNIST doesn't go beyond training and testing (validating).

Usually there is either a separate deploy prototxt, although it is possible to combine the training, testing (validation), and deployment into a single file. So yes, you would need to modify the data layer in the deploy prototxt for the specific way you want to feed in the data and you would also want need to remove the loss and accuracy layers (assuming you don't have the ground truth labels).

Jonathan

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users+unsubscribe@googlegroups.com.
To post to this group, send email to caffe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caffe-users/66273d2a-56e6-40a6-ac16-0ec85f8f98fd%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[wswfc]

Thanks a lot Jonathan. Got it!

在 2017年10月11日星期三 UTC+8下午9:28:04，Jonathan R. Williford写道：

It looks like the tutorial for MNIST doesn't go beyond training and testing (validating).

Usually there is either a separate deploy prototxt, although it is possible to combine the training, testing (validation), and deployment into a single file. So yes, you would need to modify the data layer in the deploy prototxt for the specific way you want to feed in the data and you would also want need to remove the loss and accuracy layers (assuming you don't have the ground truth labels).

Jonathan

On Wed, Oct 11, 2017 at 1:41 PM, wswfc <wsw...@gmail.com> wrote:

Hi Przemek D,

Thanks for your help. So looks like the doc in http://caffe.berkeleyvision.org/gathered/examples/mnist.html is not accurate. The training will generate .caffemodel file and to use it, we need to modify the net prototxt file to a deploy.prototxt file? Is that understanding correct?

在 2017年10月11日星期三 UTC+8下午6:45:39，Przemek D写道：

For ways to load a trained model and query it - see the examples.

W dniu środa, 11 października 2017 12:39:02 UTC+2 użytkownik wswfc napisał:

Hi,

I'm just starting getting into caffe and am walking through the tutorials on the site. Here is one question that I wish anyone could shed some lights on:

In the simplest example doc: http://caffe.berkeleyvision.org/gathered/examples/mnist.html

it says:

The final model, stored as a binary protobuf file, is stored at

lenet_iter_10000

which you can deploy as a trained model in your application, if you are training on a real-world application dataset.

But how? how can I use such model for deployment purpose? I couldn't find any doc related to it. Seems the "caffemodel file" way is deprecated now that all we got is a single file?

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users...@googlegroups.com.

To post to this group, send email to caffe...@googlegroups.com.