search

alirezatwk / Convolutional-Neural-Network-Based-Image-Watermarking-using-Discrete-Wavelet-Transform

Apache License 2.0
21 stars 2 forks source link

TRAINING ERROR #2

Closed x4jc1 closed 11 months ago

x4jc1 commented 2 years ago

hi~ here i meet the error: Input to reshape is a tensor with 265600 values, but the requested shape has 65536 could u pls explain the reasons for this?

x4jc1 commented 2 years ago

hi~ I resized coco images to size 256*256 before training. but i'm not sure if it will work.

3254657307 commented 1 year ago

Did you run it successfully?

alirezatwk commented 1 year ago

hi~ here i meet the error: Input to reshape is a tensor with 265600 values, but the requested shape has 65536 could u pls explain the reasons for this?

Hi. We hard coded the size of input image in the JPEGAttack class. However, the hard coded size is not necessary. You can change it by yourself.

Resizing the input images to 256x256 would be fine.