marcellacornia / mlnet

A Deep Multi-Level Network for Saliency Prediction. ICPR 2016
MIT License
94 stars 37 forks source link

Layer weight shape (3, 3, 640, 64) not compatible with provided weight shape (64, 3, 3, 3) #27

Open rabeya-akter opened 3 years ago

rabeya-akter commented 3 years ago

image

I'm unable to solve this problem.

James6699 commented 1 year ago

I meet the same issue.How to solve this problem?

dragonstrong commented 1 year ago

Keras uses tensorflow by default, to slove this problem you should modify the keras.json file as follows: "backend": "theano" "image_dim_ordering": "th"

foxquan commented 1 year ago

/

Keras 默认使用 tensorflow,要解决这个问题,您应该按如下方式修改 keras.json 文件: “backend”: “theano” “image_dim_ordering”: “th”

hello,l meet the same issue.I use the method you mentioned.but can't solve the problem.what the versions of tensorflow and keras you used?