EvalGEN

EvalGEN is a complementary benchmark built from synthetic images generated by modern models such as Flux, GoT, Infinity, NOVA, and OmiGen, spanning both diffusion and autoregressive architectures. It targets the robustness of AIGI detectors against the latest, rapidly evolving AIGC generators.

EvalGEN 是一款互补型基准测试集，其数据集源自 Flux、GoT、Infinity、NOVA、OmiGen 等现代模型生成的合成图像，覆盖扩散（diffusion）与自回归（autoregressive）两类模型架构。该基准测试集旨在评估AIGI检测器抵御最新且快速演进的AIGC（Artificial Intelligence Generated Content）生成器时的鲁棒性。