object_detection_sample

[Mspiz]

Hi All,

on the sample object_detection_sample I have this output on different network:

[ INFO ] InferenceEngine: 
    API version ............ 1.1
    Build .................. 11653
Parsing input parameters
[ INFO ] Loading plugin

    API version ............ 1.1
    Build .................. lnx_20180510
    Description ....... MKLDNNPlugin
[ INFO ] Loading network files:
    /opt/intel/computer_vision_sdk/deployment_tools/my_ir_model/res_mtcnn_r.xml
    /opt/intel/computer_vision_sdk/deployment_tools/my_ir_model/res_mtcnn_r.bin
[ ERROR ] Can't find output layer named bbox_pred

Any suggestion to go forward? ;)

[Monique_J_Intel · Accepted solution]

Hi Carmine,

This sample is for Faster RCNN models. So the error that you are seeing is due to the sample's expectance of an output bounding box prediction layer int the model that you are supplying. If your model is not a Faster RCNN model and doesn't have the expected output then you can use the object_detection_ssd_sample.

Kind Regards,

Monique Jones

[Mspiz]

Thanks Monique Jones.. it works fine for object_detection_ssd_sample.

Cheers

Carmine Spizuoco

[Mspiz]

Hi,

I just try  on object sample network and it work fine.

My doubt is about the creation of IR model, like explain in the guide: https://software.intel.com/en-us/articles/OpenVINO-IE-Samples

the command line is:

python3 ${MO_ROOT_PATH}/mo_Caffe*.py --input_model <path_to_model]/VGG16_faster_rcnn_final.Caffe*model --input_proto <path_to_model]/deploy.prototxt --extensions <path_to_object_detection_sample]/fasterrcnn_extensions

but the fasterrcnn_extensions file is not in the path_to_object_detection_sample.

In my optimizer transformation this parameter is not present but works fine.

Is this a good way? 

Thanks.

[Gurasz__Ky]

I have the same error (cant find bbox_pred) but I am using the faster_rcnn_resnet50_coco_2018_01_28 example, and converted to IR, trying to run it on a raspberry pi using the ncs2. Any ideas?

EDIT: Using the object_detecion_demo