IDEA-Research / MaskDINO

[CVPR 2023] Official implementation of the paper "Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentation"
Apache License 2.0
1.15k stars 102 forks source link

run demo.py for instance segmentation #79

Open homa-sotoudefard opened 1 year ago

homa-sotoudefard commented 1 year ago

I've run demo.py for inference demo with pre-trained models, I got this "Please compile MultiScaleDeformableAttention CUDA op with the following commands: cd maskdino/modeling/pixel_decoder/ops sh make.sh ". when I run "sh make.sh", I get this error "nvcc fatal : Unsupported gpu architecture 'compute_86'. the version of cuda that nvcc --version print for me is "cuda10.1". Must I install the newer version of cuda?

FengLi-ust commented 1 year ago

You can refer to deformable detr for more details about cuda installation, as we use its deformable cuda implementation.

garychan22 commented 1 year ago

@FengLi-ust the output result in opencv window is distored with input 512 * 512 by running demo.py