if i want The training input data is four channels, because my four-channel data is spliced ​​from four single-channel 2D images. How should I modify the code?

[watertianyi]