Input size different to 1x64x64x64 throws Exception

[defaultUser3214]

Hey,
I fed input tensors of e.g. the shape 1x48x48x48 into this network package. Unfortunately, I get the error message in the image.

Can anybody tell me, how the network can be adapted to also use it with custom sized squared input tensors?
Thanks a lot in advance!

[14renus]

I found that increasing size / padding the input fixed this error for me.

[Doge2408]

我发现增加大小/填充输入为我修复了这个错误。

Hello, can you tell me in detail how to change it to your own input shape? thank you very much.

[shijianjian]

I think it has to be at least 2**n (n >= 5), otherwise it will not have enough image space to perform the convolution, etc.

[Doge2408]

Thank you very much for your reply！But my data set is videos. I try to input 16 frames of pictures into the network as a group, but I don't know how to change the size or size of the convolution core.

…

------------------ Original ------------------ From: Jian Shi ***@***.***> Date: Sat,Dec 4,2021 5:23 PM To: shijianjian/EfficientNet-PyTorch-3D ***@***.***> Cc: Doge2408 ***@***.***>, Comment ***@***.***> Subject: Re: [shijianjian/EfficientNet-PyTorch-3D] Input size different to 1x64x64x64 throws Exception (#11) I think it has to be at least 2**n (n >= 5), otherwise it will not have enough image space to perform the convolution, etc. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

[shijianjian]

Say, change from Conv3D(kernel_size=(3, 3, 3)) to Conv3D(kernel_size=(1, 3, 3))will probably work for your case.

[Doge2408]

Thank you very much!!

…

------------------ Original ------------------ From: Jian Shi ***@***.***> Date: Sat,Dec 4,2021 5:39 PM To: shijianjian/EfficientNet-PyTorch-3D ***@***.***> Cc: Doge2408 ***@***.***>, Comment ***@***.***> Subject: Re: [shijianjian/EfficientNet-PyTorch-3D] Input size different to 1x64x64x64 throws Exception (#11) Say, change from Conv3D(kernel_size=(3, 3, 3)) to Conv3D(kernel_size=(1, 3, 3))will probably work for your case. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.

[riyaj8888]

its not working at all .
i changed all kernel size from (3,3,3) to (1,,3,3) but i am getting error

[darshvirbelandis]

Please see my issue posted here : #16

I am having a similar issue but I am very lost. Please help !! This EfficientNet is perfect but im just stuck on the input shape.