AdelaiDet icon indicating copy to clipboard operation
AdelaiDet copied to clipboard

“Not compiled with GPU support“ with images on README

Open Parker-Lyu opened this issue 3 years ago • 2 comments

Thanks for your amazing work! I used the image on list on README: tianzhi0549/adet:latest, but got some wrong with it. here is the message:

Traceback (most recent call last): File "demo/demo.py", line 87, in predictions, visualized_output = demo.run_on_image(img) File "/root/code/adet/demo/predictor.py", line 53, in run_on_image predictions = self.predictor(image) File "/root/code/detectron2/detectron2/engine/defaults.py", line 217, in call predictions = self.model([inputs])[0] File "/root/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 550, in call result = self.forward(*input, **kwargs) File "/root/code/adet/adet/modeling/blendmask/blendmask.py", line 115, in forward detector_results, detector_losses = self.blender( File "/root/code/adet/adet/modeling/blendmask/blender.py", line 80, in call rois = self.pooler(bases, [x.pred_boxes for x in proposals]) File "/root/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 550, in call result = self.forward(*input, **kwargs) File "/root/code/detectron2/detectron2/modeling/poolers.py", line 229, in forward return self.level_poolers[0](x[0], pooler_fmt_boxes) File "/root/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py", line 550, in call result = self.forward(*input, **kwargs) File "/root/code/detectron2/detectron2/layers/roi_align.py", line 106, in forward return roi_align( File "/root/code/detectron2/detectron2/layers/roi_align.py", line 31, in forward output = _C.roi_align_forward( RuntimeError: Not compiled with GPU support

here is my CUDA message : | NVIDIA-SMI 440.118.02 Driver Version: 440.118.02 CUDA Version: 10.2

Looking forward to your reply!

Parker-Lyu avatar Dec 16 '21 12:12 Parker-Lyu

same issue I met

aiboys avatar Dec 21 '21 06:12 aiboys

CUDA is not found when building detectron2. You should make sure python -c 'import torch; from torch.utils.cpp_extension import CUDA_HOME; print(torch.cuda.is_available(), CUDA_HOME)'

You can refer to this: https://detectron2.readthedocs.io/en/latest/tutorials/install.html#common-installation-issues

anas-zafar avatar Mar 23 '22 12:03 anas-zafar