The traditional way to detect cracks inside concretes is to use A) Radar data of the section of the concrete and B) Label Image of the section to train a pix2pix model. But, this methodology struggles to detect depending on the shape and position. So Instead of training the model with A:B = 1:1 ratio, use 2 radar data for 1 label. This is why it’s called semi-3d training. The model kind of gets to understand not only how the section itself relates to cracks, but also how the inner changes relate to inner cracks.
