Object Presence Recognition on OPOPE
77.92AccuracyOPERA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| OPERABackbone=LLaVA-1.52026.02 | 77.92 | 86.86 | 85.81 | |
| MiddleLayerBackbone=InstructBLIP2026.02 | 77.14 | 83.13 | 82.44 | |
| SPARCBackbone=LLaVA-1.52026.02 | 77.04 | 86.13 | 85.04 | |
| BaselineBackbone=LLaVA-1.52026.02 | 76.75 | 86.17 | 85.03 | |
| KVSmooth (Ours)Backbone=LLaVA-1.52026.02 | 74.6 | 91.2 | 88.9 | |
| SPARCBackbone=InstructBLIP2026.02 | 74.57 | 82.87 | 81.84 | |
| VCDBackbone=InstructBLIP2026.02 | 74.23 | 83.51 | 82.31 | |
| BaselineBackbone=InstructBLIP2026.02 | 73.94 | 83.26 | 82.06 | |
| KVSmooth (Ours)Backbone=InstructBLIP2026.02 | 73.89 | 85.19 | 83.69 | |
| PAIBackbone=InstructBLIP2026.02 | 73.69 | 82.3 | 81.19 | |
| VCDBackbone=LLaVA-1.52026.02 | 73.57 | 83.98 | 82.59 | |
| OPERABackbone=InstructBLIP2026.02 | 73.03 | 81.93 | 80.76 | |
| MiddleLayerBackbone=LLaVA-1.52026.02 | 70.97 | 89.43 | 86.51 | |
| PAIBackbone=LLaVA-1.52026.02 | 70.95 | 89.39 | 86.47 | |
| KVSmooth (Ours)Backbone=MiniGPT-42026.02 | 68.13 | 90.74 | 86.59 | |
| SPARCBackbone=MiniGPT-42026.02 | 68.08 | 88.37 | 84.72 | |
| MiddleLayerBackbone=MiniGPT-42026.02 | 68.08 | 89.61 | 85.69 | |
| BaselineBackbone=MiniGPT-42026.02 | 67.94 | 87.43 | 83.94 | |
| VCDBackbone=MiniGPT-42026.02 | 67.86 | 87.39 | 83.88 | |
| PAIBackbone=MiniGPT-42026.02 | 67.65 | 88.43 | 84.6 | |
| OPERABackbone=MiniGPT-42026.02 | 67.65 | 88.85 | 84.93 |