search

BaowenZ / RaDe-GS

RaDe-GS: Rasterizing Depth in Gaussian Splatting
Other
481 stars 27 forks source link

UnboundLocalError: local variable 'scene_info' referenced before assignment #55

Closed chojaework closed 1 month ago

chojaework commented 1 month ago

Hi, I get this error when I try to run this command.

python train.py -s <path to DTU dataset> -m <output folder> -r 2 --use_decoupled_appearance

Detailed error logs is given as below.

parser: ArgumentParser(prog='train.py', usage=None, description='Training script parameters', formatter_class=<class 'argparse.HelpFormatter'>, conflict_handler='error', add_help=True) Optimizing /RaDe-GS/outputs Output folder: /RaDe-GS/outputs [24/09 16:26:05] Tensorboard not available: not logging progress [24/09 16:26:05] Traceback (most recent call last): File "/RaDe-GS/train.py", line 306, in <module> training(dataset=lp.extract(args), File "/RaDe-GS/train.py", line 65, in training scene = Scene(dataset, gaussians) File "/RaDe-GS/scene/__init__.py", line 50, in __init__ with open(scene_info.ply_path, 'rb') as src_file, open(os.path.join(self.model_path, "input.ply") , 'wb') as dest_file: UnboundLocalError: local variable 'scene_info' referenced before assignment

Can I get some help solving this issue please?? Thank you in advance! :)

chojaework commented 1 month ago

Solved by setting the right path to the training dataset.