SeitaroShinagawa/chainer-partial_convolution_image_inpainting

Issues

• Obvious outline around the hole

  #18 opened 6 years ago by zgxiangyang
  3

• ValueError: could not broadcast input array from shape (64,3,8,8) into shape (64,3,7,7) when testing.

  #33 opened 5 years ago by greysou1
  0

• Evaluation using pre-trained .npz file

  #31 opened 5 years ago by greysou1
  3

• ZeroDivisionError: division by zero

  #32 opened 5 years ago by greysou1
  10

• test error

  #24 opened 6 years ago by erhuodaosi
  2

• would you give us a model to eval

  #2 opened 6 years ago by jianlong-yuan
  2

• The error "divsion by zero"

  #7 opened 6 years ago by sunnydai
  19

• load the npz file and then train

  #21 opened 6 years ago by yuzhesu
  2

• Little help with the requirements

  #30 opened 5 years ago by greysou1
  1

• I'm sorry to bother you again,but I have some questions in finetune

  #28 opened 6 years ago by erhuodaosi
  21

• How to print mask during one batch?

  #29 opened 5 years ago by Zhaoboling
  0

• I'm sorry to bother you,but I have some questions

  #27 opened 6 years ago by erhuodaosi
  2

• unwanted boarder around the hole due the way tv loos calculated?

  #26 opened 6 years ago by AmanuelHirpa
  3

• Replacement order

  #20 opened 6 years ago by yuzhesu
  1

• how to set the test/val_dataset and test/val_iter

  #25 opened 6 years ago by xiejiamingly
  3

• CUBLAS_STATUS_NOT_SUPPORTED

  #22 opened 6 years ago by chudonezen
  1

• the difference of the generated masks between the code and the original paper

  #23 opened 6 years ago by micklexqg
  0

• The question about the mask

  #13 opened 6 years ago by quelladai
  4

• Difference with original implementation

  #1 opened 6 years ago by liuguilin1225
  15

• Questions about training time and inpainting real images

  #14 opened 6 years ago by xhh232018
  3

• int object has no attribute shape

  #19 opened 6 years ago by yuzhesu
  3

• License?

  #15 opened 6 years ago
  2

• The problem of the edge

  #12 opened 6 years ago by yuzhesu
  5

• The test of the real broken image

  #16 opened 6 years ago by quelladai
  0

• Do you try 512 resolution？

  #3 opened 6 years ago by hudongloop
  3

• how to test on a real broken image?

  #6 opened 6 years ago by xuanzhangyang
  3

• Probable issue in calc_loss_tv with solution

  #11 opened 6 years ago by dhruva77
  3

• Perceptual loss issue

  #10 opened 6 years ago by dhruva77
  1

• about prepocessing

  #9 opened 6 years ago by syan1992
  2

• The error Unsupported dtype object

  #8 opened 6 years ago by yuzhesu
  27

• where is the upsampling

  #5 opened 6 years ago by lipanpeng
  9

• Where can I set VGG.caffemodel ?

  #4 opened 6 years ago by jeejeelee
  3