AI-Generated Image Detection on WildRF (In-the-wild) benchmark leaderboard · SOTA2 Research

AI-Generated Image Detection on WildRF (In-the-wild)

98.8Accuracy (Real)LaDeDa

Evaluation Results

Method	Links
LaDeDa2026.02		98.8	11.9	55.4
NPR2026.02		98	24.3	61.2
AIDE2026.02		97.3	19.5	58.4
CNNSpot2026.02		95.9	29	62.5
DINOv3-Linear2026.02		94.8	97.5	96.1
Gram-Net2026.02		94.7	20.5	57.6
SAFE2026.02		91.8	30.9	61.3
DDA2026.02		89.9	90.8	90.4
FreqNet2026.02		73.1	55.9	64.5
UnivFD2026.02		69.3	63.7	66.5
PE-CLIP-Linear2026.02		67.9	99.4	83.6
DINOv2-Linear2026.02		64.3	77.2	70.5
OMAT2026.02		63.3	71.5	67.4
SigLIP2-Linear2026.02		59.7	98.4	79
MetaCLIP2-Linear2026.02		47.8	97.9	72.8
MetaCLIP-Linear2026.02		46.1	92.3	69.2
SigLIP-Linear2026.02		38.3	89.7	64
Effort2026.02		17.9	95.5	56.7