One paper accepted by ECCV 2022. We propose a practical model inversion attack against deep models requiring less prior information. Looking forward to seeing you Tel Aviv in October!