Skip to main content

Show tensor as image

tip

neetbox.add_image allows you to send a tensor as image and see it in browser.

How it looks like​

image-20231220185959324

Basic Usage​

import neetbox
import torch

a = torch.zeros((300, 300))
a[:150, :150] = 1
b = torch.zeros((300, 300))
b[150:, :150] = 2
c = torch.zeros((300, 300))
c[150:, 150:] = 3
t = torch.stack([a, b, c])
neetbox.add_image(name="as CWH", image=t, dataformats="CWH")
neetbox.add_image(name="as NWH", image=t, dataformats="NWH")
t = torch.stack([t, t, t])
neetbox.add_image(name="as NCWH", image=t, dataformats="NCWH")

while True:
    pass

image-20231220221623777

tip

As you can see above, dataformats controls how neetbox understand the tensor as an image.

For example, dataformats="NCWH" means the first axis of the tensor is Number of batches, the second axis is Channel, and the third axis is Width, and the fourth axis is Height.

Pass a string consists of 'N', 'C', 'H', 'W' and neetbox will understand your tensor.

API description​

neetbox.add_image send an image to frontend display.

def add_image(name: str, image, dataformats: str = None)

Args:

  • image (Union[np.array, Image.Image, Tensor]): image from cv2 and PIL.Image as well as tensors are supported
  • name (str): name of the image, used in frontend display
  • dataformats (str): if you are passing a tensor as image, please indicate how to understand the tensor. For example, dataformats="NCWH" means the first axis of the tensor is Number of batches, the second axis is Channel, and the third axis is Width, and the fourth axis is Height.