yifanlu0227
commented
7 months ago DAIR-V2X does not need a modality assignment_file. It is typically specified in its dataset code.
Thus, I just use other dataset's assignment file (for example, v2xset_4modality.json) as a placeholder for this term when I train the models.
But it is misleading, so I change it to 'none' when releasing the code. I have to admit that this part of the code has not been tested yet, I will correct the code as soon as possible. And for now, you can fill in the assignment_path with another dataset's json file like I did before. It would not be actually used for DAIR-V2X dataset.
deyang2000
Hi , there! I had a problem when using dairv2x related files. These yaml files have assignment_path: None，when I try to use
python opencood/tools/train.py -y opencood/hypes_yaml/dairv2x/CameraOnly/camera_coalign.yaml
An error has been reported:
Dataset Building /data/cooperative-vehicle-infrastructure/train.json /data/cooperative-vehicle-infrastructure/cooperative/data_info.json None Traceback (most recent call last): File "opencood/tools/train.py", line 189, in
main()
File "opencood/tools/train.py", line 37, in main
opencood_train_dataset = build_dataset(hypes, visualize=False, train=True)
File "/home/liyf/HEAL-main/opencood/data_utils/datasets/init.py", line 27, in build_dataset
dataset = fusion_dataset_func(base_dataset_cls)(
File "/home/liyf/HEAL-main/opencood/data_utils/datasets/intermediate_heter_fusion_dataset.py", line 63, in init
else read_json(params['heter']['assignment_path'])
File "/home/liyf/HEAL-main/opencood/utils/common_utils.py", line 100, in read_json
with open(file_path, 'r') as f:
FileNotFoundError: [Errno 2] No such file or directory: 'None'
I think it is caused by the above reasons. I checked the yaml file for the V2Xset dataset and it has：
assignment_path: "opencood/logs/heter_modality_assign/v2xset_4modality.json"
How to solve this problem, thank you very much!