training of the camera-based model

[dempsey-wen]

trafficstars

Hi,

Thank you for open-sourcing your excellent work.

I tried to train the camera-based model with opencood/hypes_yaml/opv2v/CameraOnly/camera_v2xvit.yaml with only dataset path changed, but the model failed to converge.

Could you please share the details about how you train the camera-based model? Did you train the model based on the collaboration base, or did you train it from scratch?

[dempsey-wen]

Sorry for bothering, just found that I forgot to set the range in inference. After setting the range to '51.2,51.2', the result is correct now.

[yifanlu0227]

glad to hear :)

[RyanBlack5728]

抱歉打扰，刚发现我忘了设置推理范围。将范围设置为 '51.2,51.2' 后，现在结果正确。

Hello! Where should I modify the range? Thank you!

[MahatKC]

@RyanBlack5728

On the yaml file for the config you're using. Just set cav_lidar_range to [-51.2, -51.2, -3, 51.2, 51.2, 1].