Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Community Manager
441 Views

Unknown command line flag 'alsologtostderr'

I tried to run mobilenets example in ncapzoo whith description given in https://github.com/movidius/ncappzoo/blob/master/tensorflow/mobilenets/README.md.

 

It fails when i try to run "make" command with the following error:

 

` - "nception_v3_2016_08_28.tar.gz" saved [100885009/100885009]

 

(cd model/v3; tar -xvf inception_v3_2016_08_28.tar.gz;)

 

inception_v3.ckpt

 

TF_MODELS_PATH is /home/pi/Development/tensorflow/models

 

Exporting GraphDef file…

 

(cd model/v3; python3 /home/pi/Development/tensorflow/models/research/slim/export_inference_graph.py \

 

--alsologtostderr \

 

--model_name=inception_v3 \

 

--batch_size=1 \

 

--dataset_name=imagenet \

 

--image_size=299 \

 

--output_file=inception_v3.pb;)

 

FATAL Flags parsing error: Unknown command line flag 'alsologtostderr'

 

Pass --helpshort or --helpfull to see help on flags.

 

Makefile:72: recipe for target 'export' failed

 

make: *** [export] Error 1'
Tags (1)
0 Kudos
3 Replies
Highlighted
Community Manager
12 Views

Re: Unknown command line flag 'alsologtostderr'

@GoldenWings It sounds like you may be trying to run the make commands with Tensorflow models on the Raspberry Pi. This a known issue with NCSDK version 1.12. You can try generating the graph files on your PC and try running the examples using the python3 command followed by the app's python script name. Please refer to the release note # 2 on the release notes page.

0 Kudos
Highlighted
Community Manager
12 Views

Re: Unknown command line flag 'alsologtostderr'

i cant generate the graph on my pc, because i am running linux mint and when i try to generate the graph it says

 

/bin/sh: 1: mvNCProfile: not found

 

Makefile:96: recipe for target 'profile' failed

 

make: *** [profile] Error 127

 

Any recommendation?
0 Kudos
Highlighted
Community Manager
12 Views

Re: Unknown command line flag 'alsologtostderr'

@GoldenWings Maybe generate the graph files on a VM and then transfer over to the pi?

0 Kudos