hi, where is the trainning code?

[zhangyunming]

good joob ,but i have not seen the code which you said in your 'readme'.

[HYOJINPARK]

Thanks for your comments.
This work have sponsored by Naver Corporation.
Today all of the security checks are ended and I submit code.
A member from Naver will upload our code soon and I will notice the link as soon as possible.
The final code will be released in Naver Clova AI open GitHub.
I am sorry to delay and thanks for your patients.

[HYOJINPARK]

The code is released!! Thanks for your waiting.
https://github.com/clovaai/ext_portrait_segmentation

[zhangyunming]

hi, @HYOJINPARK ,thank you ,i have seen the project.
and is there a demo.py that read a pic or camera?

[HYOJINPARK]

Sorry for late answer
You can check etc/Visualize_video.py

[yasar-rehman]

Hi @HYOJINPARK,

Thank you for sharing the code.

I have tried the etc/Visualize_video.py file, however, I couldn't find the model weights "../result/Dnc_SINet11-24_2218/model_3.pth". I have also tried with the ExtremeC3.pth and SINet.pth, but it's giving me a dimension mismatch between the model and the weights. Could you guide me in solving this issue?

[HYOJINPARK]

@yasar-rehman Thanks for your asking.
The reason of mismatching dimension is with or without Lavasz loss
If you used the SIN.pth file with ./setting/Test_SINet.json
You can run it

The files are in here
https://github.com/clovaai/ext_portrait_segmentation

[yasar-rehman]

@HYOJINPARK Thank you for the help. :)