These converted images are not thermal images
ckavak opened this issue · comments
Thank you for your contribution.
Based on my 15 years experience in thermal imaging applications, unfortunately I have to say that these translated images are not thermal images. These are just inverted grayscale visible images.
Thermal image characteristics is more than converting black to white and white to black.
I aware that it is really difficult problem such that learning from visible data to thermal data during training process should avoid overfitting.
I will try to train this network with another dataset in the near future, then may share the results.