cannot make the same result as the repo example image

[jhoon-kim]

hello, first, thanks for sharing your work

i attempted to reproduce the same result as the example image from readme.md but there is a big difference in quality

i used the provided weight, data, and config should i use different parameters to reproduce the result?

i attatched the result that i made please understand that the file is compressed due to file size issue

[CHANGXU12]

I have the same blurry depth, too. Waiting for the reply. I found timm downloads tf_efficientnetv2_b3 checkpoints(tf_efficientnetv2_b3-57773f13.pth) at first. Is it different from what you used in training?

[jhoon-kim]

I have the same blurry depth, too. Waiting for the reply. I found timm downloads tf_efficientnetv2_b3 checkpoints(tf_efficientnetv2_b3-57773f13.pth) at first. Is it different from what you used in training?

i didn't train the model, used the given weight i couldn't find the file that you said, there is "model.safetensors"

windows 10 torch 1.13.1+cu117 timm 0.9.2

what version of timm do you use?

[eugenelyj]

I am using: 1) torch 1.12.0+cu113 and 2) timm 0.5.4 @CHANGXU12 @jhoon-kim Could you switch to these version and try agian?

[jhoon-kim]

I am using: 1) torch 1.12.0+cu113 and 2) timm 0.5.4 @CHANGXU12 @jhoon-kim Could you switch to these version and try agian?

I reinstalled that you said, now I am getting the similar results thanks! can you provide the "requirements.txt"?

[eugenelyj]

@jhoon-kim Sure, i will update the readme. Thanks for the suggestions!

[eugenelyj]

I have updated the installation tutorial. Feel free to ask when you have new questions.