Using mxnet pre-trained image classification model in R

Question

Running the tutorial at the link below, I get the error below when I get to the step below. No errors before this point.

http://mxnet.io/tutorials/r/classifyRealImageWithPretrainedModel.html

prob <- predict(model, X=normed) [19:01:35] D:\chhong\mxnet\dmlc-core\include\dmlc/logging.h:235: [19:01:35] d:\chhong\mxnet\src\operator./concat-inl.h:152: Check failed: (dshape[j]) == (tmp[j]) Incorrect shape[2]: (1,320,15,15). (first input shape: (1,576,14,14)) Error: InferShape Error in ch_concat_3c_chconcat: [19:01:35] d:\chhong\mxnet\src\operator./concat-inl.h:152: Check failed: (dshape[j]) == (tmp[j]) Incorrect shape[2]: (1,320,15,15). (first input shape: (1,576,14,14))

To ask a troubleshooting question you need to provide the code and data to reproduce your problem within the question itself — Hack-R, Dec 14 '16 at 00:54
I am the maintainer for the mxnet R package. First, I suggest you open issues on github. Second, the reason for your problem is that the Windows package is too old. It hasn't been updated for quite a while because I don't have access to any Windows machine now. — Qiang Kou, Dec 15 '16 at 03:43

score 0 · Answer 1 · answered Mar 02 '18 at 00:05

0

Not sure if the tutorial got fixed or one of the components got fixed but this tutorial executes without errors when testing today on Mac

answered Mar 02 '18 at 00:05

Simon Corston-Oliver

226
1
6

As mentioned in [Qiang's comment](https://stackoverflow.com/questions/41132794/using-mxnet-pre-trained-image-classification-model-in-r/49060756#comment69513882_41132794), the problem appears on WINDOWS machines. – Mogsdad Mar 02 '18 at 01:27

score 0 · Answer 2 · answered Mar 06 '18 at 01:40

I have run the tutorial on my Windows Server R2 2012 machine with mxnet version 0.10.1 and R version 3.4.3.

It runs okay, and the only thing that went wrong is that on the time I run it at the first time require(mxnet) package returned me:

> require(mxnet)
Loading required package: mxnet
Error: package or namespace load failed for ‘mxnet’:
 object ‘set_global_graph_attrs’ is not exported by 'namespace:DiagrammeR'

I had to install and load DiagrammeR manually to make sure it starts. Here is the code to do so:

# Make sure you have devtools installed already
require(devtools)
install_version("DiagrammeR", version = "0.9.0", repos = "http://cran.us.r-project.org")
require(DiagrammeR)

After that I run the tutorial again and it worked fine.

Using mxnet pre-trained image classification model in R

2 Answers2