cross_view_transformers
cross_view_transformers copied to clipboard
About test dataset
Hi, i'm training this model with Nuscenes dataset.
when I train cross-view transformer network using this code, model's best IoU for road reach 71%
but in the "Cross-view Transformers for real-time Map-view Semantic Segmentation" paper road's best IoU is 74.3%
Is the reason why my experimental results are different is that the dataset for scoring in this paper is not a validation dataset? Or is it simply because the hyperparameters to get the highest score are different?
Same here. Thank you for the great work. I simply ran trainer.validate(model_module, datamodule=data_module, ckpt_path=ckpt_path) with the checkpoint you provided (cvt_nuscenes_vehicles_50k.ckpt) but I don't see the 36.0 or 37.5 [email protected] in the paper. All I could see was 34.8. Is there a separate test set other than the validation set? Or is there a checkpoint that was learned additionally?
Hi, i'm training this model with Nuscenes dataset.
when I train cross-view transformer network using this code, model's best IoU for road reach 71%
but in the "Cross-view Transformers for real-time Map-view Semantic Segmentation" paper road's best IoU is 74.3%
Is the reason why my experimental results are different is that the dataset for scoring in this paper is not a validation dataset? Or is it simply because the hyperparameters to get the highest score are different?
SOrry to disturb you . I just run the command like " python3 train.py +experiment=cvt_nuscenes_vehicle data.dataset_dir=datasets/nuscenes data.labels_dir=datasets/cvt_labels_nuscenes" But it go with error like below: Global seed set to 2022 Error executing job with overrides: ['+experiment=cvt_nuscenes_vehicle', 'data.dataset_dir=datasets/nuscenes', 'data.labels_dir=datasets/cvt_labels_nuscenes'] Error locating target '../cross_view_transformer.model.encoder.Encoder', see chained exception above. full_key: model.encoder
Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
How can solve that? Thanks
Hi @curie3170,
I got the validation results similar to yours, do you figure out why?
Thanks.
Hello, do you have this problem?? Error executing job with overrides: ['+experiment=cvt_nuscenes_vehicle' git.exc.InvalidGitRepositoryError: /home/leezy/disk10t/lzx/cross_view_transformers
Hi, i'm training this model with Nuscenes dataset. when I train cross-view transformer network using this code, model's best IoU for road reach 71% but in the "Cross-view Transformers for real-time Map-view Semantic Segmentation" paper road's best IoU is 74.3% Is the reason why my experimental results are different is that the dataset for scoring in this paper is not a validation dataset? Or is it simply because the hyperparameters to get the highest score are different?
SOrry to disturb you . I just run the command like " python3 train.py +experiment=cvt_nuscenes_vehicle data.dataset_dir=datasets/nuscenes data.labels_dir=datasets/cvt_labels_nuscenes" But it go with error like below: Global seed set to 2022 Error executing job with overrides: ['+experiment=cvt_nuscenes_vehicle', 'data.dataset_dir=datasets/nuscenes', 'data.labels_dir=datasets/cvt_labels_nuscenes'] Error locating target '../cross_view_transformer.model.encoder.Encoder', see chained exception above. full_key: model.encoder
Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
How can solve that? Thanks
did you solve this ??
Hi @curie3170,
I got the validation results similar to yours, do you figure out why?
Thanks.
did you solve this ??
Have you had this problem before, and how did you solve it, thank you! Error executing job with overrides: ['+experiment=cvt_nuscenes_vehicle' git.exc.InvalidGitRepositoryError: /home/leezy/disk10t/lzx/cross_view_transformers
Hi @curie3170,
I got the validation results similar to yours, do you figure out why?
Thanks.
hi, i have solved this problem ,you just need to enter :git init
Hi, i'm training this model with Nuscenes dataset. when I train cross-view transformer network using this code, model's best IoU for road reach 71% but in the "Cross-view Transformers for real-time Map-view Semantic Segmentation" paper road's best IoU is 74.3% Is the reason why my experimental results are different is that the dataset for scoring in this paper is not a validation dataset? Or is it simply because the hyperparameters to get the highest score are different?
SOrry to disturb you . I just run the command like " python3 train.py +experiment=cvt_nuscenes_vehicle data.dataset_dir=datasets/nuscenes data.labels_dir=datasets/cvt_labels_nuscenes" But it go with error like below: Global seed set to 2022 Error executing job with overrides: ['+experiment=cvt_nuscenes_vehicle', 'data.dataset_dir=datasets/nuscenes', 'data.labels_dir=datasets/cvt_labels_nuscenes'] Error locating target '../cross_view_transformer.model.encoder.Encoder', see chained exception above. full_key: model.encoder
Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
How can solve that? Thanks
hi, i have solved this problem ,you just need to enter :git init