Inference demo for Detectron 2 not using any resources

I’m using Detectron2 with pretrained COCO-PanopticonSegmentation model on a GTX960. But as you can see, it’s barely using any resources. (That spike you see is me opening the screenshot utility)

Thus, the inference is taking ages, even on Colab.

What could be going wrong?
What should I read to fix it?

The Windows Task Manager doesn’t show the compute workload by default, if I’m not mistaken, so you should either select another view from the drop down menu or use nvidia-smi to check the GPU utilization.

1 Like