Fig. 2: Schematic of the process, from the 3D structure to the absorption curve output.

Image processing can predict the absorption curve from the 2D image of a given structure. A 3D structure with similar properties and different shapes can be modeled as 2D images. All the geometric properties such as the lattice constant, material type, polarization, and thickness should be the same. Under these conditions, the shape of the structure can be shown as a 2D image. This image is fed into a deep neural network. The network can predict the absorption curve from the given images. The deep neural network is a combination of a CNN and an RNN