noahzn / Lite-Mono

[CVPR2023] Lite-Mono: A Lightweight CNN and Transformer Architecture for Self-Supervised Monocular Depth Estimation
MIT License
527 stars 58 forks source link

About the test evaluation on Make3D dataset #128

Closed MrBelly closed 5 months ago

MrBelly commented 5 months ago

Dear Author,

I hope you are having a good day. and thank you for your work. I am trying to reproduce the results of make 3D dataset but the values are slightly different from the values proposed in the paper, I would be happy, if you help me with this. Thanks in advance.

my results abs_rel sq_rel rmse rmse_log 0,312 3.278 7.136 0,161

script.txt

noahzn commented 5 months ago

Hi, please take a look at this link.

MrBelly commented 5 months ago

Hi, thanks for your reply. I have already manipulated the code from the link. It gives similar results in monodepth2 but I could not get similar results for lite-mono.

noahzn commented 5 months ago

Please make sure if you are using the correct lite-mono checkpoint. What results did you get after modifying your code?

MrBelly commented 5 months ago

Ok I found out my error thanks for your time and reply