gitouni / calibnet_pytorch Goto Github PK

First of all, thank you for sharing this repo!

So I'm trying to run the test.py script.
Kitti dataset is downloaded and unzipped, first the color, then velodyne and then calib with overwriting, as described in the Dataset Preparation section.
I've activated my venv envorinment, I then ran the given python test.py ... command and iteratively pip installed everything it said my environment is lacking.
But in the end I get the following error

(venv) nofreewill@nofreesill:~/Documents/CalibNet_pytorch$ python test.py --inner_iter=1 --pretrained=./checkpoint/cam2_oneiter_best.pth --skip_frame=1 --pcd_sample=-1
args have been received, please wait for dataloader...
Traceback (most recent call last):
  File "/home/nofreewill/Documents/HomeOffice/CalibNet_pytorch/test.py", line 95, in <module>
    test_dataset = BaseKITTIDataset(args.dataset_path,args.batch_size,test_split,CONFIG['dataset']['cam_id'],
TypeError: BaseKITTIDataset.__init__() got an unexpected keyword argument 'extend_ratio'

how can i get poses in data?
Is this self-generated?

Hi,

Could you provide the pip install file? I don't think it is provided?

why i got these messages:
Ground truth poses are not avaialble for sequence 11.
Ground truth poses are not avaialble for sequence 12.
Ground truth poses are not avaialble for sequence 13.
when i run test.py?
it seems that kitti dataset dosen't have ground truth poses of sequence 11-22?