Tutorial 3 ValueError: Protocol message has no singular "pad" field.

[homer]

Hi!

I get the following error during Tutorial 3 loading a file:

Traceback (most recent call last):
  File "/usr/local/src/expresso/src/net/configViewItem.py", line 180, in onThreeDotsClickedSlot
    self.subWidget.loadTreeWidget()
  File "/usr/local/src/expresso/src/custom/netWidget.py", line 370, in loadTreeWidget
    [outputDim,bgarray]=self.addLayer(layer.name,layer.type,str(layer),layer.bottom)
  File "/usr/local/src/expresso/src/custom/netWidget.py", line 566, in addLayer
    layerdim=self.funcDict[layertype](bottomDict,layertext)
  File "/usr/local/src/expresso/src/custom/layers/calcConvolution.py", line 17, in operate
    if(conv_param.HasField("stride")):paramdict["stride"]=conv_param.stride
  File "/usr/lib/python2.7/dist-packages/google/protobuf/internal/python_message.py", line 598, in HasField
    'Protocol message has no singular "%s" field.' % field_name)
ValueError: Protocol message has no singular "stride" field.

I assume conv_param.HasField("stride") returns true but the protobuf message HasField returns false / there is no such field.

Is it a configuration problem? It is the first try after installing expresso. Tutorial 1 and 2 are doing well.

[Jaley Dholakiya]

it is due to change in caffe configuration . . . you can try with slightly older version of caffe . . .   stride has become repeated variable opposed to being an optional variable earlier . . .

thanks for reporting this issue . . .

[homer]

I tried a couple of versions, between 30.8.215 but they do not compile any more. 

Does that mean that expresso is broken? Will there be a compilable/runnable/usable version in the future or is the project dead?

Is there a more specific recommendation? What versions of the librarys do you use in the last known working install?

[homer]

Finally I got this working commenting out 3 lines. I do not know if this will lead to something, but at least I am able to follow the next tutorial steps.

[Jaley Dholakiya]

you can opt for merge request in git for contributing

[Amir Abdi]

I still haven't been able to get over this problem.
All my convolution and pooling layers have a stride value (even those with default value of 1), but the problem does not resolve.

Any suggestions?

[Michael Antolovich]

Lack of reply may mean it is dead :-(

[Michael Antolovich]

homer, which 3 lines did you comment out and did you run into any other problems 9should I give up now)?

[Farmer Lu]

i met the same problem when i installed expresso with official caffe.
later i found that expresso comes with its own caffe, and they work smoothly.

which caffe do you use ?

btw, I can run Expresso after  commeted out 3 lines ( don't know if same as he did)
but i cannot access that workstation, so i cannot tell you what lines, and in what file.

[Jaley Dholakiya]

Hey, caffe has changed significantly since expresso was released. Therefore it is highly suggested to only use the version of caffe as documented in val.serc.iisc.ernet.in/expresso .