At this line, dimensions (metos) [22,25) are selected. The low clouds dataset has only 8 metos so it will cause an error. Besides, the tensor resulting from to_0_1(input_tensor[i, 22:25]) will have a shape (3,img_width,img_height). Images in the low_clouds dataset have only 1 channel so the concatenation on the last (-1) axis will fail (cannot concat (3,x,y) with (1,x,y) on axis -1)
At this line, dimensions (metos) [22,25) are selected. The low clouds dataset has only 8 metos so it will cause an error. Besides, the tensor resulting from
to_0_1(input_tensor[i, 22:25])
will have a shape (3,img_width,img_height). Images in the low_clouds dataset have only 1 channel so the concatenation on the last (-1) axis will fail (cannot concat (3,x,y) with (1,x,y) on axis -1)