clks-wzz / PRNet-Depth-Generation

A implementaion of depth generation based on [PRNet](https://github.com/YadiraF/PRNet), which was used in the paper ***Exploiting Temporal and Depth Information for Multi-frame Face Anti-spoofing***
MIT License
95 stars 31 forks source link

Question about the depth map #5

Open taylover-pei opened 4 years ago

taylover-pei commented 4 years ago

Nice. You have done a great job. But I have a question about the depth map.

In the original paper, the values of depth map are normalized to [0:1]. But, in your code, when I printed the values of the generated depth map, I find the values are just like this: image

How should I do?

Thank you very much. Looking forward to your reply.

clks-wzz commented 4 years ago

You can normalize the depth map by dividing it with 255.0.

punitha-valli commented 4 years ago

@clks-wzz @roger-cv @taylover-pei @clks-wzz Thanks for sharing your code,

ValueError: The passed save_path is not a valid checkpoint: E:\yzu\lab\face anti spoofing\PRNet-Depth-Generation-master\Data\net-data\256_256_resfcn256_weight

but I always get this error, even when I fixed the path with my windows, it's in the same location. But its ways the path is not valid, can you please help me in this, Thanks in advance.

lalalanding commented 4 years ago

Nice! How do you make the fake depth map as zero? thank you

punitha-valli commented 4 years ago

@lalalanding

hi,

Thank you so much for your reply,

I have done with that, now fro certain images in don't get the depth map, it shows like,

Warning: No face detected

but, still, the image has a good face,

can you please tell me how to overcome it.

I have attached some images below,

image

image

image 1_1_11_3_4 1_1_12_2_5 1_2_11_3_2 2_1_14_5_2 2_2_14_1_3

roger-cv commented 4 years ago

Nice! How do you make the fake depth map as zero? thank you Can you generate fake depth map as zero? I need your help, Thank you very much.

22wei22 commented 3 years ago

@lalalanding

hi,

Thank you so much for your reply,

I have done with that, now fro certain images in don't get the depth map, it shows like,

Warning: No face detected

but, still, the image has a good face,

can you please tell me how to overcome it.

I have attached some images below,

image

image

image 1_1_11_3_4 1_1_12_2_5 1_2_11_3_2 2_1_14_5_2 2_2_14_1_3

I have the same problem. Do you solve it?

punitha-valli commented 3 years ago

I these case, the face detector said then the length will be 0

So I used opencv as a face detector

it works for me

On Tue, 3 Nov 2020 at 11:30, 22wei22 notifications@github.com wrote:

@lalalanding https://github.com/lalalanding

hi,

Thank you so much for your reply,

I have done with that, now fro certain images in don't get the depth map, it shows like,

Warning: No face detected

but, still, the image has a good face,

can you please tell me how to overcome it.

I have attached some images below,

[image: image] https://user-images.githubusercontent.com/52031723/91415975-56a06980-e881-11ea-8aa0-e3d6320f5bdd.png

[image: image] https://user-images.githubusercontent.com/52031723/91416063-720b7480-e881-11ea-938f-529d5cdc06c1.png

[image: image] https://user-images.githubusercontent.com/52031723/91416098-7f286380-e881-11ea-9538-f8facd91eaac.png [image: 1_1_11_3_4] https://user-images.githubusercontent.com/52031723/91416160-96ffe780-e881-11ea-9e1e-b2decdd79af5.jpg [image: 1_1_12_2_5] https://user-images.githubusercontent.com/52031723/91416164-98311480-e881-11ea-8363-1c5b8f6f96c8.jpg [image: 1_2_11_3_2] https://user-images.githubusercontent.com/52031723/91416168-98c9ab00-e881-11ea-9254-60c407f04b91.jpg [image: 2_1_14_5_2] https://user-images.githubusercontent.com/52031723/91416172-99624180-e881-11ea-9007-e97cc2451777.jpg [image: 2_2_14_1_3] https://user-images.githubusercontent.com/52031723/91416175-99fad800-e881-11ea-985b-732c1cf0b3c7.jpg

I have the same problem. Do you solve it?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/clks-wzz/PRNet-Depth-Generation/issues/5#issuecomment-720885588, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMM7B2YQADH5WGLZGPGJX3TSN52OZANCNFSM4JS2NQXA .