Synthetic Image Detection on UniversalFakeDetect GANs (test) benchmark leaderboard · SOTA2 Research

Synthetic Image Detection on UniversalFakeDetect GANs (test)

100ProGAN AccuracyRINE

Evaluation Results

Method	Links
RINE2026.05		100	100	99.3	100	99.6	99.9	88.9	99.4	99.5	100	99.8	100	80.6	97.9	95.4	99.6
Effort2026.05		100	100	95.1	100	99.6	99.9	99.9	97.7	99.6	100	100	100	87.6	98.9	97.4	99.5
HydraPrompt2026.05		99.9	100	96.9	100	98.8	99.8	98.2	99.9	98.4	100	100	100	93.7	97.5	98	99.6
LGrad2026.05		99.8	100	94.8	99.7	82.5	92.4	85.9	94.7	99.7	99.9	73.7	83.2	60.6	67.8	86.2	92.2
FatFormer2026.05		99.8	100	99.9	100	98.9	99.9	87.7	97.4	100	100	99.9	100	89.4	97.3	96.5	99.2
UnivFD2026.05		99.7	100	78.8	97.4	91.2	99	91.9	99.8	96.3	99.9	91.9	100	80	89.4	88.1	97.8
C2P-CLIP2026.05		99.7	100	99.4	100	95.3	99.9	90.7	99.9	95.3	100	96.6	100	89.9	97.3	95.3	99.6
F3Net2026.05		97.9	100	84.5	99.5	65.5	73.4	81.2	89.7	100	100	57	59.2	59.9	83	78.5	88.1
FreqNet2026.05		97.9	99.6	97.6	99.9	90.5	96.1	95.8	99.5	90.2	99.7	93.4	98.6	97.4	99.9	94.7	99
Frank2026.05		85.7	81.3	73.1	68.5	76.9	70.8	86.5	80.8	85	77	67.3	65.3	50.1	55.3	75	71.2
Durall2026.05		79	73.9	63.6	58.8	69.5	62.9	65.4	60.8	99.4	99.4	67	63	50.5	50.2	70.2	66.4
CNNSpot2026.05		64.6	92.7	52.8	82.8	51.6	70.5	58.6	81.5	51.2	74.3	53.6	86.6	50.6	51.5	57.3	79.6