Open hanjoonwon opened 6 months ago
In such situation you may have to reduce the rendering resolution of both image and mask. But I think 8GB seems too small for these codes. You may also need to use a much smaller SAM model for extracting masks, features and inference.
In such situation you may have to reduce the rendering resolution of both image and mask. But I think 8GB seems too small for these codes. You may also need to use a much smaller SAM model for extracting masks, features and inference.
Thankyou use smaller pretrained vit model?
In such situation you may have to reduce the rendering resolution of both image and mask. But I think 8GB seems too small for these codes. You may also need to use a much smaller SAM model for extracting masks, features and inference.
Thankyou use smaller pretrained vit model?
Yeah, but this may still not solve this problem totally.
in this block when using prompt_segmenting i got this oom error how can i avoid oom problem?? my laptop is asus and rtx 2080super with 8gb vram
window 11, wsl2 ubuntu 22.04, anaconda envirioment
Details
OutOfMemoryError Traceback (most recent call last) Cell In[15], line 7 5 img = view.original_image * 255 6 img = cv2.resize(img.permute([1,2,0]).detach().cpu().numpy().astype(np.uint8),dsize=(1024,1024),fx=1,fy=1,interpolation=cv2.INTER_LINEAR) ----> 7 predictor.set_image(img) 8 sam_feature = predictor.features 9 # sam_feature = view.original_features File ~/SuGaR/SegAnyGAussians/third_party/segment-anything/segment_anything/predictor.py:60, in SamPredictor.set_image(self, image, image_format) 57 input_image_torch = torch.as_tensor(input_image, device=self.device) 58 input_image_torch = input_image_torch.permute(2, 0, 1).contiguous()[None, :, :, :] ---> 60 self.set_torch_image(input_image_torch, image.shape[:2]) File ~/anaconda3/envs/seggau/lib/python3.9/site-packages/torch/utils/_contextlib.py:115, in context_decorator.