Hello,

Thanks for your great work. I have a question about "Calibrating from three orthogonal vanishing points". This is a classic problem and it's not the main point of your paper, but I also read about it in your blog (https://yichaozhou.com/post/20190402vanishingpoint/) and my question is, the first element (S_{11}) of the matrix S = K^-T K^-1 is 1/f^2, which will be very small when f is sufficiently large, leading to numerical instability. Do you have any idea how to handle this? Thanks in advance.

Hi, thanks for the great work.

I want to test your code, however, I'm struggling to run it. I followed your notes on installation and downloaded the pre-trained models. I'm trying to execute the following command as said in the readme:

python vectorize_u3d.py logs/pretrained-wireframe/npz/003576000 --vpdir logs/pretrained-vanishing-points/npz/000096000 57

First, I needed to figure out that I have to use python=3.7 instead of python=3.9 to resolve dependency conflicts during install.

After installation seemed fine, I'm now getting following error:

RuntimeError: module compiled against API version 0xe but this version of numpy is 0xd
Traceback (most recent call last):
  File "vectorize_u3d.py", line 39, in <module>
    from wireframe.optimize import (
  File "/home/brakebein/PycharmProjects/shapeunity/wireframe/optimize.py", line 5, in <module>
    import cvxpy as cvx

Wenn I decrease numpy version to 1.14, I get another error:

ModuleNotFoundError: No module named 'numpy.core._multiarray_umath'
Traceback (most recent call last):
  File "vectorize_u3d.py", line 39, in <module>
    from wireframe.optimize import (
  File "/home/brakebein/PycharmProjects/shapeunity/wireframe/optimize.py", line 5, in <module>
    import cvxpy as cvx

And during install I already got following message:

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
mkl-fft 1.2.0 requires numpy>=1.16, but you have numpy 1.14.6 which is incompatible.
cvxpy 1.1.17 requires numpy>=1.15, but you have numpy 1.14.6 which is incompatible.

However, increasing the numpy version to 1.16 and above throws the RuntimeError as said above.

Despite of consulting Google, stackoverflow, etc., I could not resolve this issue. So, my question is: Which dependency versions are required to successfully run your code?

Any clue that I got Segmentation fault (coredumped) when excuting python vectorize_u3d.py logs/pretrained-wireframe/npz/003576000 --vpdir logs/pretrained-vanishing-points/npz/000096000 57 output: K: 2.6102218630949925 [[ 2.6102219 0. 0. 0. ] [ 0. 2.6102219 0. 0. ] [ 0. 0. 0. -1. ] [ 0. 0. 1. 0. ]] Segmentation fault (core dumped)

Thank you very much for your research. I have downloaded your data and code, and try to run and debug it. I would like to ask how this project can extract lines, junctions, depth and vanishing points from only one image. I would appreciate it and hope you can give me some suggestions.

As titled. I followed steps in README and ran specified scripts, but it seems like the command of downloading SU3.zip doesn't work. Is it possible to update the script/command? Thank you!

Dear author，I want to ask how your training dataset is obtained through blender and sceneCity, such as how your 23000 images and jason files are obtained.Can you explain the specific steps or provide relevant codes?
Thanks!

Thanks for your good job. I have a small problem in the file of train.py and trainer.py. Could you tell me how do I use this function which is "def test(self, fname):" in trainer.py. Thanks for your again.

I have downloaded the pre-trained models and the huge dataset. Running the vectorization & visualization command with the SU3 dataset works. Now, I would like to test some of my own images with the pre-trained model to see how well it works. However, I'm not sure how to do that. How do I have to organize custom images and what command I have to run to evaluate the pre-trained model with these images?

I also tried to run the prediction command with the SU3 dataset.

python train.py --eval -d 0 -i default --from logs/pretrained-wireframe/checkpoint_latest.pth.tar logs/pretrained-wireframe/config.yaml

However, I'm getting the error that it cannot use a CUDA device. My older Nvidia graphics card is not compatible with cudatoolkit 10.x. But it looks like that the pre-trained model has somewhere saying inside that it wants to run things on CUDA, and not fall back on CPU.

Traceback (most recent call last):
  File "train.py", line 204, in 
    main()
  File "train.py", line 124, in main
    checkpoint = torch.load(resume_from)
  File "/home/brakebein/anaconda3/envs/shapeunity/lib/python3.7/site-packages/torch/serialization.py", line 608, in load
    return _legacy_load(opened_file, map_location, pickle_module, **pickle_load_args)
  File "/home/brakebein/anaconda3/envs/shapeunity/lib/python3.7/site-packages/torch/serialization.py", line 787, in _legacy_load
    result = unpickler.load()
  File "/home/brakebein/anaconda3/envs/shapeunity/lib/python3.7/site-packages/torch/serialization.py", line 743, in persistent_load
    deserialized_objects[root_key] = restore_location(obj, location)
  File "/home/brakebein/anaconda3/envs/shapeunity/lib/python3.7/site-packages/torch/serialization.py", line 175, in default_restore_location
    result = fn(storage, location)
  File "/home/brakebein/anaconda3/envs/shapeunity/lib/python3.7/site-packages/torch/serialization.py", line 151, in _cuda_deserialize
    device = validate_cuda_device(location)
  File "/home/brakebein/anaconda3/envs/shapeunity/lib/python3.7/site-packages/torch/serialization.py", line 135, in validate_cuda_device
    raise RuntimeError('Attempting to deserialize object on a CUDA '
RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

So, my next thought was that I have to train the model by myself on the CPU. So, I need to run the training command:

python ./train.py -d 0,1,2,3 --identifier baseline config/hourglass.yaml

But there is the next error I cannot resolve. I tried it several times, but it always errors at the same image or data chunk (in the folder that has been created in logs, 3 npz files has been generated, and in viz, 000002_img.jpg has been generated but not the other files (see screenshot)).

logs/.../viz/000000600/

progress | sum    | jmap   | jdir   | lmap   | joff   | ldir   | dpth   | jdep   | speed
Running validation...                                                                            Traceback (most recent call last):
  File "./train.py", line 204, in 
    main()
  File "./train.py", line 200, in main
    trainer.train()
  File "/media/brakebein/Daten4TB/projects/shapeunity/wireframe/trainer.py", line 260, in train
    self.train_epoch()
  File "/media/brakebein/Daten4TB/projects/shapeunity/wireframe/trainer.py", line 200, in train_epoch
    self.validate()
  File "/media/brakebein/Daten4TB/projects/shapeunity/wireframe/trainer.py", line 131, in validate
    self._plot_samples(i, index, H, target, f"{viz}/{index:06}")
  File "/media/brakebein/Daten4TB/projects/shapeunity/wireframe/trainer.py", line 225, in _plot_samples
    mask_target = target["jmap"][i].cpu()
IndexError: index 2 is out of bounds for dimension 0 with size 2

Of course, it would be better to avoid the training process. I just want to test some custom images on your pre-trained model, but it doesn't seem to be trivial. Any help is appreciated.

Hello, Zhou, thanks for your excellent work. I am now trying to reconstruct 3D wireframe from real-world images and therefore I think L3D dataset will be helpful. Would you please plan to release the L3D dataset?