chenhsuanlin / bundle-adjusting-NeRF

BARF: Bundle-Adjusting Neural Radiance Fields 🤮 (ICCV 2021 oral)
MIT License
793 stars 114 forks source link

RuntimeError: shape '[1, 3, 160000]' is invalid for input of size 0 #91

Open jiangyijin opened 8 months ago

jiangyijin commented 8 months ago

Hello, thank you very much for your work. I encountered this issue when reconstructing my dataset. I used single-channel black and white images. TRAINING START validating: 0%| | 0/4 [00:00<?, ?it/s]C:\ProgramData\anaconda3\envs\barf-env\Lib\site-packages\torch\functional.py:507: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at C:\cb\pytorch_1000000000000\work\aten\src\ATen\native\TensorShape.cpp:3550.) return _VF.meshgrid(tensors, kwargs) # type: ignore[attr-defined] Traceback (most recent call last): File "D:\3dgs\gaosi\bundle-adjusting-NeRF-main\bundle-adjusting-NeRF-main\train.py", line 33, in main() File "D:\3dgs\gaosi\bundle-adjusting-NeRF-main\bundle-adjusting-NeRF-main\train.py", line 30, in main m.train(opt) File "D:\3dgs\gaosi\bundle-adjusting-NeRF-main\bundle-adjusting-NeRF-main\model\nerf.py", line 54, in train if self.iter_start==0: self.validate(opt,0) ^^^^^^^^^^^^^^^^^^^^ File "C:\ProgramData\anaconda3\envs\barf-env\Lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, *kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "D:\3dgs\gaosi\bundle-adjusting-NeRF-main\bundle-adjusting-NeRF-main\model\barf.py", line 66, in validate super().validate(opt,ep=ep) File "C:\ProgramData\anaconda3\envs\barf-env\Lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(args, kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "D:\3dgs\gaosi\bundle-adjusting-NeRF-main\bundle-adjusting-NeRF-main\model\base.py", line 153, in validate loss = self.graph.compute_loss(opt,var,mode="val") ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "D:\3dgs\gaosi\bundle-adjusting-NeRF-main\bundle-adjusting-NeRF-main\model\nerf.py", line 218, in compute_loss image = var.image.view(batch_size,3,opt.H*opt.W).permute(0,2,1) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ RuntimeError: shape '[1, 3, 160000]' is invalid for input of size 0