TrustCLIP: Learning Private Visual Features via Adversarial Reconstruction figure
AlphaXiv 中文概览(可滚动查看)