Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.

Can I use my own ssd model with NCSDK ?

idata
Employee
474 Views

hi, I used my own ssd model using codes from apps/video_object, but executed failed. I changed class number with my class number, mask, etc. Its execution result is as below:

 

python video_objects.py actual video resolution: 320.0 x 240.0 [ nan nan nan ..., nan nan nan] Traceback (most recent call last): File "video_objects.py", line 414, in <module> sys.exit(main()) File "video_objects.py", line 377, in main run_inference(display_image, ssd_mobilenet_graph) File "video_objects.py", line 238, in run_inference num_valid_boxes = int(output[0]) ValueError: cannot convert float NaN to integer

 

Thanks in advance. :smile:

0 Kudos
2 Replies
idata
Employee
295 Views

What modifications should I do if using my own ssd model? :wink:

0 Kudos
idata
Employee
295 Views

I also get all Nan result when port my similar ssd detect net. as my last two lay like:

 

layer {

 

name: "Concat_LocConfpreds"

 

type: "Concat"

 

bottom: "Concat_Locpreds"

 

bottom: "Concat_Confpreds"

 

top: "Concat_LocConfpreds"

 

concat_param {

 

axis: 2

 

}

 

}

 

layer {

 

name: "last_layer_reshape"

 

type: "Reshape"

 

bottom: "Concat_LocConfpreds"

 

top: "last_layer_reshape"

 

reshape_param {

 

shape {

 

dim: 0

 

dim: 0

 

dim: 0

 

}

 

}

 

}
0 Kudos
Reply