umarspa / hairanalysis Goto Github PK

Hi @UmarSpa , thank you for sharing.
According to the paper, the hair segmentation results can be classified into 7 hairstyle classes, but I can't find it in this repository.

When I run the code I get the following result, but it doesn't seem to match the paper result. Any idea what I could do wrong ?
Thanks

Hello,

First, just huge thanks about showing your project to us ! Great work, and i hope it will inspire many of us to make greats things !

Just wanted to tell you, that i could not test your project because, there is no tutorial so i decided to install my self things to execute your main.py, but impossible to get caffe lib on my iMac 27" but AMD GC... What am i suppose to do, to test your project ?

Thanks !

hi

when doing "overlapping 25x25 patch extraction from uncertain region", the program doesn't go any further and I have to kill the process manually.

I use ubuntu 18.04, python3 to run the program

Hi,
Instructions says you need to download model_caffenet.caffemodel and place it in Tools/CaffeNet/.
But the link is not working.
I would really appreciate if you could help.
Thanks

Hello Umar,

Firstly, thank you for the code. I am a student working on 3d reconstruction of hair from single view images. While I am not an expert in machine learning but I have to admit that running your code and getting the results was easy and charm. Yet to run through your code and understanding the paper in parallel is something I am progressing slowly :-).

I have few questions though, after running your code there are results inside the Output folder. And the results are

• An image showing the hair region as a binary mask
• An image showing the non-hair region as a binary mark
• An Image showing the hair region probability
• A numpy truth table of shape equal to image size to imply the hair regions in the pixels

Can you tell me how to detect the classification of hair style in the images? Is that part included in the code or is it something to be implemented as another layer?

Regards,

#0K

i can`t download the model because of the link error.
if someone have it let me know and can u send to me?
or have the link updated.
very thanks!

name 'caffe' is not defined how to fix theme?

Hi, I am interested in your paper and work. However, I cannot find any code related to hair classification that has been described in your paper. Could you please share any code about this part?

Thanks your project.
I run the main.py and occur an error:

i want to segmentation hair from human image in IOS app.
so i tried convert model_caffenet.caffemodel to coreml .
but this model output size is 1000

http://www.eecs.qmul.ac.uk/~urm30/HairMaterial/model_caffenet.caffemodel