search

TassnimDardouri / FCNN_LS

1 stars 0 forks source link

Where to do the inference? #1

Open Xinyue-Katrina-LI opened 2 years ago

Xinyue-Katrina-LI commented 2 years ago

Dear Dardouri,

I wonder if you could provide some hints on how to do inference on your model, i.e encoding and decoding? Which part of your code should I use?

We are the group from University of New South Wales, Sydney, doing related work to yours. We are considering citing and comparing your method with ours in a journal paper. It would be great if we could know your inference steps, so as to generate the RD curves based on our testing datasets.

Regards, Xinyue

TassnimDardouri commented 2 years ago

Dear Xinyue,

Sure, let me get the code.

Kind regards,

Message ID: @.*** -- Tassnim Dardouri Doctorante à L2TI, Université Sorbonne Paris Nord en collaboration avec CVN, CentraleSupélec

Xinyue-Katrina-LI commented 2 years ago

Dear Dardouri,

Can I ask you for a favour if it is convenient for you? Is it possible for you to test your method with Kodak dataset (with only 24 images), at bit-rate 0.2, 0.4, 0.6, 0.8, 1.0? I only need yours with PSNR and MS-SSIM values for comparison.

Thanks very much for you assistance in advance. Look forward to your reply.

Best Regards, Xinyue

TassnimDardouri commented 2 years ago

Sure, no problem. I just need some time. Which method do you need?

On Mon, Aug 8, 2022 at 11:15 AM Katrina-li-code @.***> wrote:

Dear Dardouri,

Can I ask you for a favour if it is convenient for you? Is it possible for you to test your method with Kodak dataset (with only 24 images), at bit-rate 0.2, 0.4, 0.6, 0.8, 1.0? I only need yours with PSNR and MS-SSIM values for comparison.

Thanks very much for you assistance in advance. Look forward to your reply.

Best Regards, Xinyue

— Reply to this email directly, view it on GitHub https://github.com/TassnimDardouri/FCNN_LS/issues/1#issuecomment-1207870868, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJVYFEJCITBMZT67TMHNLNDVYDFZZANCNFSM553OI7SQ . You are receiving this because you commented.Message ID: @.***>

-- Tassnim Dardouri Doctorante à L2TI, Université Sorbonne Paris Nord en collaboration avec CVN, CentraleSupélec

Xinyue-Katrina-LI commented 2 years ago

Thanks very much for you help, and please take your time.

I just need whatever the best performance is at your side. If I am not wrong, it should be the 'H-FCNN-LS' method as denoted in your journal paper 'Dynamic Neural Network for Lossy-to-Lossless Image Coding'.

TassnimDardouri commented 2 years ago

Okey great. I'll get back to you as soon as possible.

On Mon, Aug 8, 2022 at 12:10 PM Katrina-li-code @.***> wrote:

Thanks very much for you help, and please take your time.

I just need whatever the best performance is at your side. If I am not wrong, it should be the 'H-FCNN-LS' method as denoted in your journal paper 'Dynamic Neural Network for Lossy-to-Lossless Image Coding'.

— Reply to this email directly, view it on GitHub https://github.com/TassnimDardouri/FCNN_LS/issues/1#issuecomment-1207926651, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJVYFENAT376B6CQIOFPI7TVYDMIRANCNFSM553OI7SQ . You are receiving this because you commented.Message ID: @.***>

-- Tassnim Dardouri Doctorante à L2TI, Université Sorbonne Paris Nord en collaboration avec CVN, CentraleSupélec

Xinyue-Katrina-LI commented 2 years ago

Thank you so very much for doing this for us :)

Xinyue-Katrina-LI commented 2 years ago

Dear Dardouri,

Hope this message finds you well. Have you got any PSNR or MS-SSIM results on your method with the grayscale Kodak dataset?

Regards, Xinyue

TassnimDardouri commented 2 years ago

Dear Xinyue,

There is a problem with Matlab on my computer, and I couldn't run the inference. However, I have PSNR or MS-SSIM results for rates [0.07,0.1,0.15,0.2,0.25,0.3,0.4,0.5] bpp.

MS-SSIM: [0.84558852, 0.88771041, 0.92018377, 0.9396054 , 0.95100191, 0.96028986, 0.97206402, 0.97921714] PSNR: [23.84793864, 25.39402584, 27.03960012, 28.32207864, 29.38411129, 30.2735586 , 31.78531751, 33.04147171]

If you need results for the rest of the bitrates, you can use the code attached below (along with the GitHub code).

Please let me know if there is any problem with the code or if there is a file missing.

Kind regards, encode_decode.zip https://drive.google.com/file/d/14D46KHgN5mK3UVdt8fVxsg_XRxNWDV37/view?usp=drive_web

On Tue, Aug 16, 2022 at 2:56 AM Katrina-li-code @.***> wrote:

Dear Dardouri,

Hope this message finds you well. Have you got any PSNR or MS-SSIM results on your method with the grayscale Kodak dataset?

Regards, Xinyue

— Reply to this email directly, view it on GitHub https://github.com/TassnimDardouri/FCNN_LS/issues/1#issuecomment-1216027725, or unsubscribe https://github.com/notifications/unsubscribe-auth/AJVYFEKYGHO6FPRHDPVHKYLVZLRNLANCNFSM553OI7SQ . You are receiving this because you commented.Message ID: @.***>

-- Tassnim Dardouri Doctorante à L2TI, Université Sorbonne Paris Nord en collaboration avec CVN, CentraleSupélec