Open linleyzoe opened 1 year ago
Do you know how to do with it? I also have some questions!
hello, you can find split_data.json files(train.json, val.json) in this zip file https://github.com/yifanlu0227/CoAlign.
Actually, you can also find it on the official github page of dair-v2x. https://github.com/AIR-THU/DAIR-V2X/blob/main/data/split_datas/cooperative-split-data.json In this case, you can use the following code to see how they split the dataset into training and validation sets:
info_json_path = r"E:\python_work_d\project_samples\DAIR-V2X-main\data\split_datas"
info_name = "cooperative-split-data.json"
with open(os.path.join(info_json_path, info_name), 'rb') as f:
split_data = json.load(f)
split_info = split_data["cooperative_split"]["train"]
for i in range(10):
print(split_info[i])
Then, to prepare the dataset for model training, following these steps:
Can you run this program normally?
yes,can't u?
yes, I can run the code now. But how to set communication volume like papers?
yes, I can run the code now. But how to set communication volume like papers? Have you solved this problem? Do you use image in the code or only use point cloud?Thank you!
And how does his communication round K reflect
Google Drive Share Link from CoAlign: https://drive.google.com/file/d/13g3APNeHBVjPcF-nTuUoNOSGyTzdfnUK/view?usp=sharing
Hi, thanks for your inspiring work!
There seems no instructions for converting the raw DAIR-V2X to the specific training format, which is referred as
root_dir: "/GPFS/rhome/quanhaoli/workspace/dataset/my_dair_v2x/v2x_c/cooperative-vehicle-infrastructure/train.json"
in the config file.Any advice would be greatly appreciated!