Using the DeViT model to perform inference for traffic sign recognition

[bigz321]

Hello, I am trying to perform inference on traffic signs, such as trafficcones, cars, and trucks, but I'm not sure if I'm on the right track. My settings as following:

def main(
        config_file="configs/open-vocabulary/coco/vitl.yaml",
        rpn_config_file="configs/RPN/mask_rcnn_R_50_C4_1x_ovd_FSD.yaml",
        model_path="weights/trained/open-vocabulary/coco/vitl_0064999.pth",

        image_dir = '/root/src/CODA/base-val-1500/images', 
        output_dir='demo/output', 
        category_space="demo/normalized_prototypes.pth",
        device='cpu',
        overlapping_mode=True,
        topk=1,
        output_pth=False,
        threshold=0.45
    ): 

In my opinion, this COCO checkpoint can detect objects like cars and trucks. What I need to do is generate the category_space.pth file, is that correct？