Thanks for reaching out to us.
Could you please provide us the links and steps you have followed for running the experiment so that we can try it from our end.
So far, we are not at liberty to share the code outside. Apology for that.
Basically, our program is using OpenCV's UI backend support ( re-run cmake or configure script in function cvShowImage · Issue #8537 · opencv/opencv · GitHub ), whereas DevCloud's OpenCV is built without it.
Not sure if there a suggested approach in such scenario where we would still like to use OpenCV in its full capacity and be able to utilize DevCloud for our model's training and inferences?
Since the code is not available to replicate, kindly check the below suggestions.
Please refer the below link.
Try to create a conda environment by cloning the root.
conda create -n root_clone --clone base
source activate root_clone
Try to install the packages you need inside the conda environment.
Kindly let us know if this solves your issue.