Figure 6
From: Real-time coherent diffraction inversion using deep generative networks
Activation maps of select convolutional layers. (A) Input diffraction patterns. (B) Average activation of the 2nd convolutional layer of sCDI NN. (C) Average activation of the 4th convolutional layer of sCDI NN. (D) Average activation of the 2nd convolutional layer of pCDI NN. (E) Average activation of the 4th convolutional layer of pCDI NN. Successive layers of CDI NN learn features at higher Q (higher resolution). Intensity units are arbitrary.
