Classid Labels

[JVond]

New to the compute stick.  I've been using the google coral and jetson nano.

I was running the alexnet and googlenet caffe model and got outputs with no labels.  Is there a list somewhere or a way to input labels?

googlenet

classid probability
------- -----------
616     0.4355762  
50      0.1069827  
836     0.0557207  
36      0.0534245  
429     0.0518655  
494     0.0363212  
452     0.0321387  
544     0.0257031  
600     0.0205605  
772     0.0144706 

alexnet

classid probability
------- -----------
88      0.9810570  
12      0.0157526  
7       0.0009286  
90      0.0009172  
14      0.0005626  
136     0.0002139  
92      0.0001659  
129     0.0001552  
91      0.0000463  
17      0.0000393  

[HemanthKum_G_Intel]

Hi James,

You should be able to search the internet for the associated labels and annotation files. But labels and annotation are not associated with the models, they are associated with the dataset which was used to train the models, for instance, Coco, ImageNet, etc...

Well, OpenVINO doesn't ship with labels file for all supported models, so, for example, you can pull the ILSVRC2012 sysnset_words.txt from http://dl.caffe.berkeleyvision.org/caffe_ilsvrc12.tar.gz and rename it as googlenet-v1.labels. Note that the ILSVRC2012 synset_words.txt won't work for GoogLeNet V2 since V2 was trained on ILSVRC2015, you'd have to fetch the 2015 labels file when running GoogLeNet V2.

You have to run the classification sample with --lables option followed by the lables file downloaded from respective sites.