GAN-generated image detection on GANGen-Detection benchmark leaderboard · SOTA2 Research

GAN-generated image detection on GANGen-Detection

100BEGAN AccuracyLadeda

Evaluation Results

Method	Links
Ladeda2025.01		100	100	99	99.9	97.9	99.9	99	99.9	100	100	99.8	100	80.5	84.9	100	100	98.4	99.9	97.2	98.3
FatFormer2025.01		99.9	100	98.3	99.9	98.4	100	98.4	100	99.5	100	98.8	99.8	99	100	99.3	100	98.4	100	98.9	100
NPR2025.01		99.6	99.9	93.3	97.3	98.5	98.3	98.5	98.3	99.8	100	99.6	100	79.8	78.9	92.5	98.6	91.5	97.2	94.8	96.5
FreqNet2025.01		98.8	100	85.4	90.4	95.2	98.2	95.2	98.2	100	100	98.8	100	88.3	94.3	89.8	98.8	94	97.5	94	97.5
Ours2025.01		98.6	99.7	96.8	99.6	98	99.7	98.7	99.8	99	99.9	99.1	100	88.2	91.1	99	99.8	97.2	99.7	97.3	98.8
Patchfor2025.01		97.1	100	97.6	99.8	97.8	99.9	97.9	100	99.6	100	92.7	99.8	66.8	68.8	68	92.9	93.6	98.6	90.1	95.4
C2P-CLIP2025.01		94.9	100	98.4	99.9	98.4	100	98.4	100	92	99.8	97.6	99.6	99	100	90.4	99.8	98.4	100	96.4	99.9
UnivFD2025.01		88.9	96.3	87.7	96.5	89.8	99.1	89.7	99	94	98.1	82.6	91.5	94.7	98.9	90	96.8	87.4	96.5	89.4	97
F3Net2025.01		87.1	97.5	51.6	93.6	89.5	99.8	73.7	99.6	98.8	100	60.3	99.9	65.4	70	85.2	94.8	67.1	83.1	75.4	93.1
LGrad2025.01		69.9	89.2	78	87.4	50.3	50.3	57.5	67.3	89.1	99.1	54.8	68	78.5	86	68.6	93.8	71.1	82	68.6	80.8
GANDetection2025.01		67.9	100	66.7	90.6	67.8	99.7	67.7	99.3	60.9	86.2	69.6	97.2	69.6	83.5	57.4	75.1	67.6	92.4	66.1	91.6
AIDE2025.01		57.9	79.5	76.1	79.2	82	88.1	82.2	88.3	95.2	96.5	92.7	98.1	69	78.8	96.6	98.4	80.7	82.8	81.4	87.8
CNN-Spot2025.01		50.2	44.9	62.7	90.4	81.5	97.5	72.9	94.4	53.3	82.1	63	92.7	55.2	66.1	51.1	83.7	71.1	94.7	62.3	82.9