search

brandleyzhou / DIFFNet

[BMVC 2021] ''Self-Supervised Monocular Depth Estimation with Internal Feature Fusion''
109 stars 21 forks source link

Environment #8

Closed seb-le closed 2 years ago

seb-le commented 2 years ago

Hi, thank you for sharing your nice work.

Could you share the environment setting such as versions of packages for this work?

I cannot reproduce the results of this paper even if using the pretrained model that is provided in this repo.

In my evaluation: 0.1024 0.7632 4.482 1.799 0.8954 0.9645 0.9831

In the paper: 0.102 0.764 4.483 0.180 0.896 0.965 0.983

brandleyzhou commented 2 years ago

Can you show your reproduced results? my torch version is 1.1.0. I think the reason why your evaluation of the trained model is different from the paper is that the printing format.

seb-le commented 2 years ago

As your mentioned, I tried to evaluate the official model of this repo on torch 1.1.0. But, I still faced different results from the paper.

Below is the results by evaluation code of this repo: image

More precisely: 0.10242 0.76323 4.48215 0.17993 0.89543 0.96449 0.98312

Thank you.

brandleyzhou commented 2 years ago

Sorry for replying to you late. I have retrained and evaluated a model and what I got is image The model named diffnet_640x192 is available on google drive.

seb-le commented 2 years ago

Thank you for your kindness!

I downloaded and evaluated it.

Although there are some differences between your test and my test, it is very minor.

image

Again, I appreciate your kindness,

and I have really learned a lot thanks to your source code and your paper.

Thanks.