国科大杭高院发布突破性成果 应用前景很广

Core Insights - The research team led by Professor Chen Luonan from the University of Chinese Academy of Sciences and Shanghai Jiao Tong University has developed the Difface AI model, which can generate 3D facial images directly from DNA sequences, marking a significant breakthrough in 3D facial reconstruction technology [2][3] - This technology, referred to as "genetic portraiture," is expected to have important applications in fields such as criminal investigation, historical anthropology, and personalized medicine [2] Group 1 - The Difface model combines pre-trained Transformer architecture with spiral convolution technology, effectively acting as a "spatial genetic decoder" and "3D visual modeling instrument" [2] - The model enhances the matching accuracy between DNA sequences and 3D faces by approximately 12%, maintaining high generation precision even with limited DNA data [3] - The technology allows for the simulation of facial changes over time, providing new avenues for locating missing persons by generating current appearances from childhood DNA [3] Group 2 - In the field of criminal investigation, this technology could potentially solve long-standing cases by reconstructing suspect faces from biological evidence found at crime scenes [3] - In anthropology, ancient human DNA may be used to recreate the appearances of ancestors [3] - In the cosmetic medical field, it could enable personalized predictions of facial features based on genetic information [3]