whats the meaning of the function named add_depth_channels(image_tensor) in phalanx folds

[l475139105]

why can add_depth_channels function change GRAYSCALE images to three channels?And whats the utilize of the func in the processing of training model?
thx

[l475139105]

<def add_depth_channels(image_tensor):
_, h, w = image_tensor.size()
image = torch.zeros([3, h, w])
image[0] = image_tensor
for row, const in enumerate(np.linspace(0, 1, h)):
image[1, row, :] = const
image[2] = image[0] * image[1]
return image>

[l475139105]

def add_depth_channels(image_tensor): _, h, w = image_tensor.size() image = torch.zeros([3, h, w]) image[0] = image_tensor for row, const in enumerate(np.linspace(0, 1, h)): image[1, row, :] = const image[2] = image[0] * image[1] return image

[Interesting6]

Same question.
@phalanx-hk @ybabakhin
looking forward your reply.