test error about qcnet HOT 19 OPEN

@ZikangZhou Thank you very much. I have run successfully and the result on the leaderboard is:

A little bit worse than your submission.

from qcnet.

It seems that you haven’t installed the av2 api. Could you confirm whether the av2 package has been installed in your current environment?

from qcnet.

Yes, I have installed the av2 api and there is still this error message.

from qcnet.

Could you "import av2" successfully?

from qcnet.

I was facing the same issue. This got resolved using "pip install av2".

from qcnet.

@ZikangZhou Thank you. I have solved the error I mentioned and run test on a 24GB card, but there is a new error:

@shahaamirbader Did you have the same problem?

from qcnet.

Hi I haven't yet faced this error since presently I am training this model. I may face this issue during test as you are facing. By the way how long did it take for you to train the network? I have 2 GPUs , 32 GB and it's been two days and only 7 epochs. Do you know if the author has shared the checkpoints of the retrained models?

from qcnet.

@ZikangZhou Thank you. I have solved the error I mentioned and run test on a 24GB card, but there is a new error: @shahaamirbader Did you have the same problem?

This is because you're using an old version of av2-api. To avoid re-installing av2, you can try to replace the content in lib/python3.8/site-packages/av2/datasets/motion_forecasting/eval/submission.py with the latest code in https://github.com/argoverse/av2-api/blob/main/src/av2/datasets/motion_forecasting/eval/submission.py. Sorry for the inconvenience in installation.

from qcnet.

Hi I haven't yet faced this error since presently I am training this model. I may face this issue during test as you are facing. By the way how long did it take for you to train the network? I have 2 GPUs , 32 GB and it's been two days and only 7 epochs. Do you know if the author has shared the checkpoints of the retrained models?

The checkpoint can be downloaded here: https://drive.google.com/file/d/1OKBytt6N6BdRa9FWmS7F1-YvF0YectBv/view?usp=drive_link

from qcnet.

@bixiang BTW, can you tell me the test set performance of the checkpoint if you can submit it to the leaderboard?

from qcnet.

@ZikangZhou Thank you very much. I have run successfully and the result on the leaderboard is:

A little bit worse than your submission.

Thx! Actually this performance is better，because my previous submission has used ensembling (combination of 5 single models).

from qcnet.

@ZikangZhou Thank you very much. I have run successfully and the result on the leaderboard is: A little bit worse than your submission.

I submit the "submission.parquet" file into the argoverse2 eval website, but it takes a long time to get the eval result, and i haven't get the eval result until now, so what's wrong with me?

from qcnet.

@ZikangZhou Hi, Zhou! What do you mean about ensembling (combination of 5 single models)? Are those models all QCNet, or are QCNet and other models?

from qcnet.

@ZikangZhou Hi, Zhou! What do you mean about ensembling (combination of 5 single models)? Are those models all QCNet, or are QCNet and other models?

All are QCNet.

from qcnet.

@ZikangZhou Hello, Zhou. How can the 30 trajectories generated by the 5 models be integrated into trajectories for the 6 modes?

from qcnet.

@ZikangZhou Hello, Zhou. How can the 30 trajectories generated by the 5 models be integrated into trajectories for the 6 modes?

@ZikangZhou I am also confused how the 5 models are integrated.

from qcnet.

@ZikangZhou Hello, Zhou. How can the 30 trajectories generated by the 5 models be integrated into trajectories for the 6 modes?

@ZikangZhou I am also confused how the 5 models are integrated.

@rongxiaoqu do you have any idea about integrating 5 models now?

from qcnet.

@ZikangZhou Hi, Zhou! What do you mean about ensembling (combination of 5 single models)? Are those models all QCNet, or are QCNet and other models?

All are QCNet.

Hello, may I ask how much K you selected when using K-MEANS?

from qcnet.

Hi, your work is great. I try to generate the prediction results on the test set by following your code:

python test.py --model QCNet --root /path/to/dataset_root/ --ckpt_path /path/to/your_checkpoint.ckpt

But there is some error like this:

Global seed set to 2023
Processing...
  0%|          | 0/24984 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "test.py", line 40, in <module>
    test_dataset = {
  File "/content/QCNet/datasets/argoverse_v2_dataset.py", line 150, in __init__
    super(ArgoverseV2Dataset, self).__init__(root=root, transform=transform, pre_transform=None, pre_filter=None)
  File "/usr/local/envs/QCNet/lib/python3.8/site-packages/torch_geometric/data/dataset.py", line 97, in __init__
    self._process()
  File "/content/QCNet/datasets/argoverse_v2_dataset.py", line 534, in _process
    self.process()
  File "/content/QCNet/datasets/argoverse_v2_dataset.py", line 190, in process
    map_data = read_json_file(map_path)
TypeError: object() takes no arguments

Wondering what's the reason and how to deal with it. Thank you very much.

在我的例子里，当我注释掉QCNet/datasets/argoverse_v2_dataset.py 34~43行的try...except，发现环境中opencv 的依赖库（libgthread-2.0.so.0,libGL.so.0）没有正确安装。安装后问题解决。

from qcnet.