BEVFormer
BEVFormer copied to clipboard
How to train a model with different dataset?
Thanks for the best work!
When I was learning this project, I was thinking how to use different dataset for training at the same time? Different dataset means different number of cameras, different bev size. I'm out of ideas, do you have any good ideas?
Looking forward to your reply!
I also encountered the same problem. May I ask how to train BEVFormer on the Kitti dataset?
have you trained bevformer on different datasets?
I also encountered the same problem. May I ask how to train BEVFormer on the Kitti dataset?
I haven't used the Kitti dataset before. This is how I solved the two problems:
- Different number of cameras: padding zero image make all dataset have same input size.
- Different bev size: take the largest bev size in the dataset.
have you trained bevformer on different datasets?
This is how I solved the two problems: Different number of cameras: padding zero image make all dataset have same input size. Different bev size: take the largest bev size in the dataset.
I also encountered the same problem. May I ask how to train BEVFormer on the Kitti dataset?
Can you tell me how you use the Kitti dataset
you can modify the data from kitti to nuscenes type(such as sample.json,sample_data.json, and so on.) or modify dataloader.whatever,if you can create right train.pkl,val.pkl,test.pkl.you can use kitti. ---- Replied Message ---- From @.> Date 05/09/2024 16:26 To @.> Cc @.>@.> Subject Re: [fundamentalvision/BEVFormer] How to train a model with different dataset? (Issue #141) I also encountered the same problem. May I ask how to train BEVFormer on the Kitti dataset? Can you tell me how you use the Kitti dataset — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>