pkuxmq/Invertible-Image-Rescaling

Issues

• 关于PSNR和SSIM的测量问题

  #52 opened 5 months ago by ROTTK
  0

• 关于哈尔小波下采样

  #40 opened 2 years ago by zhh99
  2

• Testing with desktop captured image, I got a blur output.

  #39 opened 2 years ago by steven8274
  24

• question about image rescaling problem

  #48 opened a year ago by Feynman1999
  0

• about training convergence speed

  #51 opened a year ago by Feynman1999
  0

• The training is unstable

  #35 opened 3 years ago by shengkelong
  2

• what is train_IRN_x2_finetune.yml used for?

  #50 opened a year ago by Feynman1999
  0

• Signal high-frequency information recovery

  #49 opened a year ago by Jiayu668
  0

• H264 compression

  #44 opened 2 years ago by ld-xy
  0

• why using gt size 144?

  #47 opened a year ago by junhyoung-ew
  1

• 您好，请问该怎么做使test.py得到的fake_H图片与原GT图大小一致？

  #46 opened a year ago by dmndxld
  4

• ssim计算问题

  #45 opened 2 years ago by kerryhhh
  1

• train IRN_x4??

  #43 opened 2 years ago by ld-xy
  0

• Questions about network with jpeg

  #42 opened 2 years ago by brighteast
  4

• 图像大小的问题

  #15 opened 4 years ago by Fiona2016Xue
  5

• Questions about the models

  #41 opened 2 years ago by mrgreen3325
  1

• Training time?

  #32 opened 3 years ago by Wenchao-Du
  6

• Weird artifacts when testing on Kodak dataset

  #13 opened 4 years ago by amnesiack
  6

• About Training Details

  #38 opened 3 years ago by zhuyeye
  0

• How IRN deals with the loss caused by video compression artifact?

  #6 opened 4 years ago by coolKeen
  7

• What if the log determinant of Jacobian diverge to nan?

  #37 opened 3 years ago by Wangbk-dl
  0

• about test execute time

  #36 opened 3 years ago by Young-Zzz
  1

• Link of pretrained model is broken

  #34 opened 3 years ago by KangHuaLi
  2

• Why the PSNR value decreases by several DBs when I introduces H.264 encoding losses?

  #33 opened 3 years ago by xiaoxiongli
  1

• why is so slow? in one V100 GPU, 10 second per frame?

  #30 opened 3 years ago by xiaoxiongli
  2

• Separate downscaling from upscaling

  #31 opened 3 years ago by felipeinagaki
  2

• 如何输入LQ，得到HQ

  #29 opened 3 years ago by sym1998
  1

• Dosen't work with LQ mode

  #4 opened 4 years ago by ajuanzju
  10

• Performance on Bicubic-based LR images to HR images

  #11 opened 4 years ago by zhangdan8962
  7

• Problem about loss_ce

  #14 opened 4 years ago by MC-E
  9

• Questions regarding the the test function at IRN_model.py

  #28 opened 4 years ago by peterhan91
  2

• The formula in paper seems not the same shown in code？

  #27 opened 4 years ago by JuZiSYJ
  4

• About Gaussian Meaning

  #26 opened 4 years ago by delldu
  2

• What is the reason that let you use gradient clipping here?

  #24 opened 4 years ago by johnnylu305
  3

• About hyper-parameters lambda

  #19 opened 4 years ago by codyshen0000
  5

• A minor mistake?

  #23 opened 4 years ago by plutoyuxie
  1

• How to calculate the jacobian

  #5 opened 4 years ago by zwb0
  7

• Some questions about the experiment？

  #21 opened 4 years ago by toddwyl
  1

• Training Details

  #7 opened 4 years ago by wenbinTse
  5

• About loss_ce?

  #22 opened 4 years ago by codyshen0000
  1

• About Distribution Loss

  #20 opened 4 years ago by codyshen0000
  2

• Why we need to crop border?

  #18 opened 4 years ago by johnnylu305
  1

• about trainning by Y channel with yuv files

  #17 opened 4 years ago by flyBird-445
  5

• Where is the implementation for stage 2 with full distribution matching loss `L_distr`

  #16 opened 4 years ago by JingyunLiang
  5

• model output is the same shape as input...

  #12 opened 4 years ago by zhangqizky
  3

• 8 times downscale

  #10 opened 4 years ago by volcanoxs
  3

• Is this training process correct?

  #9 opened 4 years ago by coolKeen
  2

• GPU memory requirement

  #8 opened 4 years ago by sunbingfeng
  2

• Running on Windows 10 getting RuntimeError

  #3 opened 4 years ago by surfingnirvana
  1

• Why we need the HR reconstruction loss?

  #2 opened 4 years ago by a514514772
  2