Error in e329a2f79be94ce289a8002f1f174340.glb (Objaverse dataset)

[Fangkang515]

Thanks for your work.

I found that this code will go wrong in some scenarios, such as e329a2f79be94ce289a8002f1f174340.glb in Objaverse dataset.

ValueError: Trainer: Found array with 0 sample(s) (shape=(0, 1)) while a minimum of 1 is required by QuantileTransformer.

Could you please test it out? I am still unable to precisely locate where the problem lies.

Thank you so much~

[jian-77]

Your work is very inspiring, but I also had this issue on every attempt. which happens on
File "/home/ryan/Segmentation/SAMPart3D/exp/sampart3d/knight/code/pointcept/engines/train.py", line 381, in _get_quantile_func quantile_transformer = quantile_transformer.fit(scales.reshape(-1, 1))
Does anyone know how to fix it? Appreciate that.

[eiiijiiiy]

In my case, this is due to the out of memory issue of GPU, making SAM producing no masks. This leads to an empty scales array for quantile_transformer.fit(scales.reshape(-1, 1)).

[JarneVdH]

I had the same issue as OP and this was caused by a compatibility issue between torch and torchvision. This caused the SAM call to fail, where the error was invisible at first because of the general try-except block here:

SAMPart3D/pointcept/datasets/dataset_render_16views.py

Lines 173 to 178 in 6a508d1

	try:
	masks = SAM_model(img, points_per_side=32, pred_iou_thresh=0.9, stability_score_thresh=0.9)
	masks = masks['masks']
	masks = sorted(masks, key=lambda x: x.sum())
	except:
	masks = []

Upon inspecting the error I saw the following:

Error processing frame_0: Couldn't load custom C++ ops. This can happen if your PyTorch and torchvision versions are incompatible, or if you had errors while compiling torchvision from source. For further information on the compatible versions, check https://github.com/pytorch/vision#installation for the compatibility matrix. Please check your PyTorch version with torch.__version__ and your torchvision version with torchvision.__version__ and verify if they are compatible, and if not please reinstall torchvision so that it matches your PyTorch install.

After re-installing the environment with special care for some packages, it worked.