Hengyue-Bi
Hengyue-Bi
> def __len__(self): > return self.tensor.shape[0] Thanks for your valuable comment which truly helps me.
> def __len__(self): > return self.tensor.shape[0] Thanks for your valuable comment which truly helps me.
> > > when i run main function where is model.py i get an error,for example > > > " > > > energy = content_content + content_position > >...
[[array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)], [array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)], [array([0., 0., 0., 0., 0., 0., 0., 0.,...
Why the processed location of charbboes all zero? I am looking forward to your answering our questions, thank u!
global_encoder { backbone = resnet34 pretrained = True num_layers = 4 }
Hi, i met the same issues, and i have solved. You can use `print(self.base_path)` before line 29 in File "/home/hua/pixel-nerf/src/data/SRNDataset.py", and check the direction compared with the your dataset location.
I am trying to finetune dust3r to my own datasets. The loss functions are adopted from the original implementation. If you want to finetune dust3r, you should prepara image pairs...
> Thank you for your reply. I started training with the MV-DUSt3R model. However, I found that although the loss functions of DUSt3R and MV-DUSt3R are similar in the paper,...
> Hello, I'm trying to train DUST3R on my own dataset, too. But the result in predicting model and global-alignment will be strange. Could you talk about how you preprocess...