How to recover the data shape using Conv2DTranspose, after convolution with stride = 2, kernel_size = 3, padding =1?

Hi everyone,
My input data is (1, 3, 13, 13), after a convolution (kernel size = 3, stride = 2, padding 1), the data is (1, 32, 7, 7).

Now, I want to upsample the data back to (1, 3, 13, 13), I tried to use Conv2DTranspose( kernel_size 4, stride = 2, padding = 1), I got a data (1, 3, 14, 14). So I cannot concat the upsampled data with the input data because their dimensions of last two axes are different.

Any advices? thanks