Object Hallucination on POPE Adversarial v1.0

89.26AccuracyActive-Look

Evaluation results

Method
Active-Look2026.04	89.26	89.09	87.85	88.46
Active-Look2026.04	87.97	88.36	85.76	87.04
Active-Look2026.04	85.73	92.81	77.47	84.45
Highlight2026.04	85.34	85.79	84.71	85.25
Active-Look2026.04	85	84.28	80.4	82.28
Zoom-in2026.04	84.42	84.92	84.12	84.52
ONLY + VDC2025.12	84.4	85.78	82.47	84.09
Baseline2026.04	84.32	84.78	83.37	84.07
ONLY2025.12	83.77	85.1	81.87	83.45
VCD + VDC2025.12	83.73	88.98	77	82.56
Baseline2026.04	83.73	83.23	82.17	82.69
M3ID + VDC2025.12	83.4	91.34	73.8	81.64
Vanilla + VDC2025.12	83.2	91.09	73.6	81.42
VCD2025.12	83.1	88.46	76.13	81.83
M3ID2025.12	83.03	91.12	73.2	81.18
Active-Look2026.04	82.78	83.64	78.32	80.89
Vanilla2025.12	82.53	90.8	72.4	80.56
Baseline2026.04	81.85	88.19	72.64	79.66
Active-Look2026.04	81.25	85.93	74.86	80.01
Baseline2026.04	80.6	85.92	73.2	79.05
Active-Look2026.04	80.55	82.41	79.29	80.82
OPERA2026.04	79.89	85.12	73.04	78.62
ONLY + VDC2025.12	79.67	75.23	88.47	81.31
ONLY2025.12	79.43	75.07	88.13	81.08
VAF2026.04	78.96	81.88	74.95	78.26
Zoom-in2026.04	78.43	81.13	78.32	79.7
Highlight2026.04	78.21	80.82	77.36	79.05
VCD2026.04	78.12	80.86	73.15	76.81
Baseline2026.04	78.12	83.27	73.34	77.98
Baseline2026.04	77.33	81.68	70.47	75.66
M3ID + VDC2025.12	76.87	72.05	87.8	79.15
Vanilla + VDC2025.12	76.63	72.01	87.13	78.85
M3ID2025.12	76.63	71.79	87.73	78.97
ONLY + VDC2025.12	76.4	72.02	86.33	78.53
VCD + VDC2025.12	76.3	72.82	83.93	77.98
M3ID + VDC2025.12	76.3	72.2	85.53	78.3
Baseline2026.04	76.1	78.47	71.93	75.06
Vanilla2025.12	75.87	71.18	86.93	78.27
VCD2025.12	75.8	72.29	83.67	77.56
VCD + VDC2025.12	75.67	70.52	88.2	78.38
ONLY2025.12	75.63	71.28	85.87	77.9
M3ID2025.12	75.4	71.26	85.13	77.58
VCD2025.12	75.23	70.23	87.6	77.96
Vanilla + VDC2025.12	74.97	70.96	84.53	77.15
Vanilla2025.12	74.17	70.04	84.47	76.58