How to use dino_vit_base_weight in objectdetection?

[dbsdmlgus50]

Checklist

1. I have searched related issues but cannot get the expected help.
2. I have read the FAQ documentation but cannot get the expected help.

Thank you for devlope mmserise!!
I would like to bring the official weight of the models using the vit backbone network like Dino and proceed with the detection, but I am leaving an inquiry because there is a problem.

I want to learn fast_rcn detector by taking a bit-based weight, how do I modify the backbone?

When setting up the model

# model settings

norm_cfg = dict(type='BN', requires_grad=False)
model = dict(
type='FasterRCNN',
backbone=dict(
type='vit', #ViT enabled
arch='base',
img_size=224,
patch_size=16,
in_channels=3,
embed_dims=192,
num_layers=12,
num_heads=3,
mlp_ratio=4,
qkv_bias=True,
drop_rate=0.0,
attn_drop_rate=0.0,
norm_layer=dict(type='LN', eps=1e-6),
with_cls_token=True,
with_pos_embed=True),
neck=dict(
type='FPN',
in_channels=[192, 384, 768, 1536],
out_channels=256,
num_outs=5),

Then, I imported the fast_rcnn below. Is there anyone who can help me?

plus how to use official weight?? like movo1(official git)