Gao, Ruiyuan
Gao, Ruiyuan
大屏幕也没法满屏显示，而且某些分辨率下图标会错位。 手机屏幕这个兼容性就更差了。。。
I do not think it is trivial to support multi-batch training, right now the video training only supports one video on one process. I sometimes use the "batch" dim as...
You should find the parameter count in your training log. If not, I will check them later.
The generalization ability of camera params is limited, which is a known issue as presented in our work, MagicDrive3D. However, we have tried some parameters different from the nuScene and...
Please consider this, zero init should help to stable the training process. > Besides, if you think camera pose embedding is the key reason. In our latest work, we implemented...
I think you are talking about the Fourier Embedding, which we borrowed from NeRF. You can find the citation (Mildenhall et al., 2020) in our paper.
Please install mmdet3d from `third-party/bevfusion` if you need it.
Maybe you can try this: https://huggingface.co/botp/stable-diffusion-v1-5
I do not have time to verify this link. If anyone finds the link useful, please leave a message to let others know. Thanks.
Besides changing the image resolution, we changed the conv module for bev map. I will get back to you today. If not, please remind me.