search

JUGGHM / PENet_ICRA2021

ICRA 2021 "Towards Precise and Efficient Image Guided Depth Completion"
MIT License
323 stars 46 forks source link

Ran PENet pre-trained model and results do not match Kitti benchmark depth completion page #31

Open manifischer opened 2 years ago

manifischer commented 2 years ago

Thank you very much for a very interesting paper. I have run the PENet pre-trained model you've provided in evaluation mode on the cropped image. in the results the code provided (val.csv under results) I got RMSE=757.197 MAE=209.001, compared to RMSE=730.08 MAE=210.55 as it appears in the Kitti benchmark page. IIs there a different PENet model that matches the submitted results? or there is something in the parameters that I put wrong (I kept the parameters as is in this repository). Thanks a lot, Mani

JUGGHM commented 2 years ago

Thanks for your interest! It's the same model as it is benchmarked on test set while validated on val set. You could also refer to this issue.

manifischer commented 2 years ago

Thanks a lot, missed the issue you referred to

kingLCH commented 2 years ago

how to use penet pre-trained model,why it name is end with .tar it seems cant be loaded by "checkpoint = torch.load(args.evaluate, map_location=device)"

JUGGHM commented 2 years ago

how to use penet pre-trained model,why it name is end with .tar it seems cant be loaded by "checkpoint = torch.load(args.evaluate, map_location=device)"

Thanks for your interest! I am afraid taht it is an inherited bug and you can fix it yourself.