Different settings of classes on DAIR-V2X-I between baseline and BEVHeight

[rbldrbld]

Hello, I noticed that the experimental settings for object categories provided by the DAIR-V2X baseline are different from those used by BEVHight.

In DAIR-V2X for ImvoxelNet, only Truck, Van, Bus are mapped to 'Car'
https://github.com/AIR-THU/DAIR-V2X/blob/c65a55617f7d0a9b78dc9d107370c95bcac55dca/tools/dataset_converter/gen_kitti/label_json2kitti.py#L39

In BEVHeight, not only Truck, Van, Bus are mapped to 'Car', the Tricyclist, Motorcyclist, Barrowlist are also mapped to 'Cyclist'

BEVHeight/scripts/data_converter/gen_kitti/label_json2kitti.py

Line 40 in f5894ab

def rewrite_txt(path):

But in the paper, these two models, ImvoxelNet and BEVHeight are compared together. May I ask if this difference will affect the experimental results?