Figure 1
From: Reconstruction of Iberian ceramic potteries using generative adversarial networks
Overview of the proposed approach. (A) IberianGAN architecture. The G(x) generator is based on an encoder-decoder architecture. Upon receiving a fragment of pottery, the encoder transforms it into a vector and then the decoder generates the missing or unknown fragment. The discriminator D(x) receives the complete profile to determine if it is true or false. (B) Criteria for profile partitioning into rim and base of profiles. (C) Examples of IberianGAN generated samples from fragments for both open and closed shapes (shown in lighter color). (D) Semi-landmark analysis and RMSE values as comparing actual and artificially generated samples.
