cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Highlighted
Contributor
Contributor
453 Views
Registered: ‎07-15-2019

Decent-cpu running error

Jump to solution

 

when I run decent-cpu on my ubuntu pc, it reports:

 

decent-cpu: error while loading shared libraries: libopencv_highgui.so.2.4: cannot open shared object file: No such file or directory

 

does decent-cpu only support opencv2.4?

 

0 Kudos
1 Solution

Accepted Solutions
Highlighted
Xilinx Employee
Xilinx Employee
421 Views
Registered: ‎05-24-2019

@phoenixmy 

There are some OpenCV functions which are not backward compatible.  Precisely how that may impact DNNDK I am not certain.  Are you certain that you are not already running 2.4? You can double-check to see if the files are installed:

 

sudo locate libopencv_core.so.2.4

 

Also, you can confirm which version you have installed:

 

dpkg -l | grep libopencv

 

You could potentially setup an Anaconda virtual environment, specifically for DNNDK.  I would actually recommend doing this as you can then have multiple environments for different versions of TF, DNNDK, etc.

Sudo ld_config may also help, depending on the exact nature of your situation:

https://github.com/pjreddie/darknet/issues/382

--Quenton

 

 

View solution in original post

0 Kudos
3 Replies
Highlighted
Xilinx Employee
Xilinx Employee
422 Views
Registered: ‎05-24-2019

@phoenixmy 

There are some OpenCV functions which are not backward compatible.  Precisely how that may impact DNNDK I am not certain.  Are you certain that you are not already running 2.4? You can double-check to see if the files are installed:

 

sudo locate libopencv_core.so.2.4

 

Also, you can confirm which version you have installed:

 

dpkg -l | grep libopencv

 

You could potentially setup an Anaconda virtual environment, specifically for DNNDK.  I would actually recommend doing this as you can then have multiple environments for different versions of TF, DNNDK, etc.

Sudo ld_config may also help, depending on the exact nature of your situation:

https://github.com/pjreddie/darknet/issues/382

--Quenton

 

 

View solution in original post

0 Kudos
Highlighted
Contributor
Contributor
410 Views
Registered: ‎07-15-2019

@quentonh  已写:

@phoenixmy 

There are some OpenCV functions which are not backward compatible.  Precisely how that may impact Are you certain that you are not already running 2.4.  You can double-check to see if the files are installed:

 

sudo locate libopencv_core.so.2.4

 

Also, you can confirm which version you have installed:

 

dpkg -l | grep libopencv

 

You could potentially setup an Anaconda virtual environment, specifically for DNNDK.  I would actually recommend doing this as you can then have multiple environments for different versions of TF, DNNDK, etc.

Sudo ld_config may also help, depending on the exact nature of your situation:

https://github.com/pjreddie/darknet/issues/382

--Quenton

 

 


 

@quentonh

Thanks for your reply.

The issue is fixed when I install opencv 2.4.9. 

But when I run yolo2caffe example(https://github.com/Xilinx/Edge-AI-Platform-Tutorials/tree/master/docs/Darknet-Caffe-Conversion) and excute "bash 2_quantize.sh", 'decent-cpu' crash, below is the log("log.txt"), do you know the root cause for it?

 

 

Highlighted
Contributor
Contributor
394 Views
Registered: ‎07-15-2019

@phoenixmy  已写:

@quentonh  已写:

@phoenixmy 

There are some OpenCV functions which are not backward compatible.  Precisely how that may impact Are you certain that you are not already running 2.4.  You can double-check to see if the files are installed:

 

sudo locate libopencv_core.so.2.4

 

Also, you can confirm which version you have installed:

 

dpkg -l | grep libopencv

 

You could potentially setup an Anaconda virtual environment, specifically for DNNDK.  I would actually recommend doing this as you can then have multiple environments for different versions of TF, DNNDK, etc.

Sudo ld_config may also help, depending on the exact nature of your situation:

https://github.com/pjreddie/darknet/issues/382

--Quenton

 

 


 

@quentonh

Thanks for your reply.

The issue is fixed when I install opencv 2.4.9. 

But when I run yolo2caffe example(https://github.com/Xilinx/Edge-AI-Platform-Tutorials/tree/master/docs/Darknet-Caffe-Conversion) and excute "bash 2_quantize.sh", 'decent-cpu' crash, below is the log("log.txt"), do you know the root cause for it?

 

 


New update:

problem solved, it my mistake, I used the wrong v3.prototext file(the one for 80 classes), but acutally, my own has only 10 classes.

so the filters value in layer81 is 45, not 255.

0 Kudos