Intrinsic Morphological Relationship Guided 3D Craniofacial Reconstruction Using Siamese Cycle Attention GAN
Published in SIGGRAPH Asia 2024, 2024
Abstract: Craniofacial reconstruction is essential in forensic science and has widespread applications. It is challenging due to the detailed facial geometry, complex skull topology, and nonlinear skull-face relationship. We propose a novel approach for 3D craniofacial reconstruction using a Siamese cycle attention mechanism within Generative Adversarial Networks (GAN). Benefiting from the cycle attention mechanism, our method focuses on high-frequency features and morphological connections between the skull and face. Additionally, a Siamese network preserves its identity consistently. Extensive experiments demonstrate superior accuracy and high-quality details of our approach.
Recommended citation: Junli Zhao, Chengyuan Wang, Yu-Hui Wen, Fuqing Duan, Ran Yi, Yong-Jin Liu*, Qingdong Long, Zhenkuan Pan, and Xianfeng Gu. 2024. Intrinsic Morphological Relationship Guided 3D Craniofacial Reconstruction Using Siamese Cycle Attention GAN. In SIGGRAPH Asia 2024 Technical Communications (SA ‘24).