Open 05063112lcs opened 4 months ago
Hi, thanks for your interest. Before evaluating on Tanks and Temples (TNT) benchmark, we finetune our DTU trained model on the BlendedMVS dataset for the adaptation to the real-world scenes. This is a common practice in the MVS field. The command can be found here.
Thank you very much for your reply. Do you mean that the pre-trained model of the TNT dataset can be obtained through the training script of BlendMVS?
---Original--- From: "Tianqi @.> Date: Thu, Jun 6, 2024 23:17 PM To: @.>; Cc: @.**@.>; Subject: Re: [TQTQliu/ET-MVSNet] Testing on Tanks and Temples (Issue #14)
Hi, thanks for your interest. Before evaluating on Tanks and Temples (TNT) benchmark, we finetune our DTU trained model on the BlendedMVS dataset for the adaptation to the real-world scenes. This is a common practice in the MVS field. The command can be found here.
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>
Yes, just set BLD_CKPT_FILE
to the DTU trianed model and run the BlendedMVS training script.
I understand that for your article, you trained on DTU data set, tested on DTU data set,BlendedMVS and TNT data set, and then obtained the effect in the chart in the paper. Well, I'm replicating your work
---Original--- From: "Tianqi @.> Date: Thu, Jun 6, 2024 23:31 PM To: @.>; Cc: @.**@.>; Subject: Re: [TQTQliu/ET-MVSNet] Testing on Tanks and Temples (Issue #14)
Yes, just set BLD_CKPT_FILE to the DTU trianed model and run the BlendedMVS training script.
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>
Thanks! To be precise, we train model on the dtu training set from scratch and then test it on the dtu test set. We then fine-tune the trained model on the blendedmvs dataset and test it on TNT.
Thank you for your trouble. Good luck with your studies
---Original--- From: "Tianqi @.> Date: Thu, Jun 6, 2024 23:47 PM To: @.>; Cc: @.**@.>; Subject: Re: [TQTQliu/ET-MVSNet] Testing on Tanks and Temples (Issue #14)
Thanks! To be precise, we train model on the dtu training set from scratch and then test it on the dtu test set. We then fine-tune the trained model on the blendedmvs dataset and test it on TNT.
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>
Hello, thank you very much for your excellent work, I would like to know, for the TNT dataset, how do you get your own pre-trained model, or is it just training on the DTU dataset and testing on the rest of the dataset.