Visual Question Answering on Infrared-COCO (test)
71.38Answer CorrectnessClean
Evaluation Results
| Method | Links | |
|---|---|---|
| CleanImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.6 (13B)2026.04 | 71.38 | |
| CleanImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.5 (7B)2026.04 | 68.21 | |
| CleanImage Encoder=EVA-CLIP ViT-G/14, Models=Blip-2 FlanT5XL (4.1B)2026.04 | 65.83 | |
| CleanImage Encoder=EVA-CLIP ViT-G/14, Models=InstructBLIP FlanT5XL (4.1B)2026.04 | 64.78 | |
| QR codeImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.6 (13B)2026.04 | 64.23 | |
| QR codeImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.5 (7B)2026.04 | 63.34 | |
| CleanImage Encoder=OpenAI CLIP ViT-L/14, Models=Blip-2 FlanT5XL ViT-L (3.4B)2026.04 | 62.32 | |
| HCBImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.6 (13B)2026.04 | 61.65 | |
| HCBImage Encoder=EVA-CLIP ViT-G/14, Models=Blip-2 FlanT5XL (4.1B)2026.04 | 60.56 | |
| AdvGridImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.6 (13B)2026.04 | 59.36 | |
| AdvGridImage Encoder=EVA-CLIP ViT-G/14, Models=Blip-2 FlanT5XL (4.1B)2026.04 | 59.34 | |
| QR codeImage Encoder=EVA-CLIP ViT-G/14, Models=InstructBLIP FlanT5XL (4.1B)2026.04 | 59.1 | |
| HCBImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.5 (7B)2026.04 | 58.56 | |
| QR codeImage Encoder=OpenAI CLIP ViT-L/14, Models=Blip-2 FlanT5XL ViT-L (3.4B)2026.04 | 58.12 | |
| HCBImage Encoder=OpenAI CLIP ViT-L/14, Models=Blip-2 FlanT5XL ViT-L (3.4B)2026.04 | 57.78 | |
| QR codeImage Encoder=EVA-CLIP ViT-G/14, Models=Blip-2 FlanT5XL (4.1B)2026.04 | 57.58 | |
| HCBImage Encoder=EVA-CLIP ViT-G/14, Models=InstructBLIP FlanT5XL (4.1B)2026.04 | 57.39 | |
| AdvGridImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.5 (7B)2026.04 | 57.28 | |
| AdvGridImage Encoder=OpenAI CLIP ViT-L/14, Models=Blip-2 FlanT5XL ViT-L (3.4B)2026.04 | 55.78 | |
| UCGPImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.6 (13B)2026.04 | 55.23 | |
| AdvGridImage Encoder=EVA-CLIP ViT-G/14, Models=InstructBLIP FlanT5XL (4.1B)2026.04 | 55.21 | |
| CleanImage Encoder=OpenAI CLIP ViT-L/14, Models=OpenFlamingo (3B)2026.04 | 51.56 | |
| UCGPImage Encoder=OpenAI CLIP ViT-L/14, Models=LLaVA-1.5 (7B)2026.04 | 51.34 | |
| UCGPImage Encoder=EVA-CLIP ViT-G/14, Models=Blip-2 FlanT5XL (4.1B)2026.04 | 47.34 | |
| UCGPImage Encoder=OpenAI CLIP ViT-L/14, Models=Blip-2 FlanT5XL ViT-L (3.4B)2026.04 | 46.78 | |
| UCGPImage Encoder=EVA-CLIP ViT-G/14, Models=InstructBLIP FlanT5XL (4.1B)2026.04 | 42.23 | |
| QR codeImage Encoder=OpenAI CLIP ViT-L/14, Models=OpenFlamingo (3B)2026.04 | 41.85 | |
| AdvGridImage Encoder=OpenAI CLIP ViT-L/14, Models=OpenFlamingo (3B)2026.04 | 37.46 | |
| HCBImage Encoder=OpenAI CLIP ViT-L/14, Models=OpenFlamingo (3B)2026.04 | 35.74 | |
| UCGPImage Encoder=OpenAI CLIP ViT-L/14, Models=OpenFlamingo (3B)2026.04 | 32.03 |