AI-generated video frame detection on GenVidBench frame-level samples benchmark leaderboard · SOTA2 Research

AI-generated video frame detection on GenVidBench frame-level samples

95.5CogVideo AccuracyDTS-Det

Evaluation Results

Method	Links
DTS-Det2026.07		95.5	95.7	79.1	76.2	95.1	95.2	76.2	71.8	83.2	81.5	76.1	71.7	92.2	92.6	91.5	92.1	88.5	89	92.1	92.5	91	91.6	87.3	86.4
SigLIP2+DINOv32026.07		83.5	81.8	58.5	36.6	84	82.4	53	21.6	66.3	52.8	50.5	13.9	80.5	77.9	85.5	84.5	77	72.9	86.5	85.6	86.4	85.6	73.8	63.2
UnivFD2026.07		81	83	66.5	65.6	75.5	76.9	52	42.1	69.4	69	53	44	69	70.4	77.5	80.2	73	75.2	76.5	79.1	69	70.4	69.3	68.7
DINOv3-LoRA2026.07		76.5	80.9	61.5	75.5	57.5	81	57.5	58.3	54.6	54.8	56	57.3	77	81.3	75.5	79.8	75	79.3	76.2	80.3	76.8	81.3	67.6	73.6
F3Net2026.07		57.5	57.7	52	51.9	58.5	57.8	49	47.4	48.4	47.6	53.5	52.7	56	56.4	56.5	56.7	54	54.9	56.2	55.6	53.8	53.5	54.1	53.8
AIDE2026.07		55.8	65.1	52.2	60.2	61.5	67.3	53	61.7	47.9	55.6	58.2	67.5	58.5	73.7	58.4	63.7	52	59.6	62.5	68.1	54	58.8	55.8	63.6
ResNet502026.07		53.5	52.3	58.5	58.7	61.5	61.7	54	53.5	50.5	49.2	56.5	55.8	58	58.4	62.5	63.7	57	56.7	61.5	62.8	61.5	61.7	57.7	57.6
FreqNet2026.07		53	54.3	50	50.4	43.5	43.2	46.5	46.7	43.8	43.2	51	53.3	49.5	49.2	50	48.9	49.5	48.7	47	46.5	51	51.5	48.6	48.7
Noiseprint++2026.07		53	67.8	52.5	67.4	47.5	62.6	51	65.9	51.1	65.9	53	67.8	49	65.5	50	66.4	49.2	65.4	47.5	64.1	48.5	65.1	50.2	65.8
SigLIP2-LoRA2026.07		50.5	61.7	48.2	59	44.5	55	47.5	58.4	40.3	49.7	48.3	59.6	56.5	65	50.5	58	60.5	69.2	50.7	58.2	56	66.7	50.3	60.1
NPR2026.07		49.5	50.2	58	59.9	58.5	61	52.5	54.9	54.5	55.2	52	55.6	66.5	68.2	61.5	63.2	55	57.5	56.2	56.9	63.8	66.3	57.1	58.9
Xception2026.07		49	49.5	56.2	57.2	54.8	55	53	54.3	49.4	48.7	58	59.9	61.5	62.8	58	59.2	57	59.1	56.5	56.3	63.5	65.4	56.1	57
C2P-CLIP2026.07		49	59.8	46	56.4	48.5	59.3	49.5	60.4	49.5	60.2	47.5	58.2	52	63.1	40.5	49.8	53	64.1	38	46.5	47.2	57.6	47.3	57.7