Then, your transpose should convert a now [channel, height, width] tensor to a [height, width, channel] one. To do this, use np.transpose(image.numpy(), (1, 2, ...
確定! 回上一頁