2
$\begingroup$

Context: I'm trying to improve a pose estimation model so that it works better when my camera is in Infrared mode. Unfortunately I only have RGB images to train on.

I realize that you can't convert RGB to IR directly, but my hypothesis is that converting the RGB images to look more like IR, and then training on a dataset of combined RGB and IR images, will lead to better performance.

Are there any libraries that have tried to implement a function like this? I'm essentially looking for a function that something like this ("IR effect") - http://funny.pho.to/infrared/

$\endgroup$
1
  • $\begingroup$ What exactly do you want to achieve? I'm not expert on imaging, but isn't it that a camera either is able to record in infrared, or isn't? If your images are not registered in infrared, then you don't have such information. So do you want the images to "look like" made in infrared, while being meaningless, or do you want to guess what could be "hotter" in infrared? $\endgroup$
    – Tim
    Commented Feb 13, 2018 at 10:41

1 Answer 1

1
$\begingroup$

If you find yourself a big enough dataset of same images both in visible spectrum and infrared spectrum, then you could use a convolutional neural network for this. Input would be an RGB image, and output would be infrared version of it. In theory it might work.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.