Image Retrieval from Contextual Descriptions on IMAGECODE All (test)
90.8AccuracyHuman
Evaluation Results
| Method | Links | |
|---|---|---|
| HumanEvaluation Protocol=Zero-shot2024.05 | 90.8 | |
| BLIP-2Params=1.2B, Evaluation Protocol=Zero-shot2024.05 | 79.2 | |
| ContextBLIPParams=240M, Evaluation Protocol=Fine-tuned2024.05 | 78.5 | |
| BLIPParams=223M, Evaluation Protocol=Fine-tuned2024.05 | 77.4 | |
| ContextBLIPParams=240M, Evaluation Protocol=Zero-shot2024.05 | 77.1 | |
| BLIPParams=223M, Evaluation Protocol=Zero-shot2024.05 | 74.4 | |
| ALBEFEvaluation Protocol=Zero-shot2024.05 | 73.3 | |
| NDCRParams=440M, Evaluation Protocol=Fine-tuned2024.05 | 64.3 | |
| CLIPParams=473M, Evaluation Protocol=Fine-tuned2024.05 | 59.8 | |
| OFAEvaluation Protocol=Fine-tuned2024.05 | 52.1 | |
| UNITEREvaluation Protocol=Fine-tuned2024.05 | 50.5 | |
| ViBERTEvaluation Protocol=Fine-tuned2024.05 | 49.3 | |
| CLIPParams=473M, Evaluation Protocol=Zero-shot2024.05 | 47.8 | |
| UNITEREvaluation Protocol=Zero-shot2024.05 | 42.9 | |
| ViBERTEvaluation Protocol=Zero-shot2024.05 | 40.8 | |
| ContextBLIPParams=240M, Evaluation Protocol=Fine-tuned2024.05 | 35.7 | |
| BLIPParams=223M, Evaluation Protocol=Fine-tuned2024.05 | 34.1 | |
| NDCRParams=440M, Evaluation Protocol=Fine-tuned2024.05 | 34.1 | |
| ContextBLIPParams=240M, Evaluation Protocol=Zero-shot2024.05 | 31 | |
| CLIPParams=473M, Evaluation Protocol=Fine-tuned2024.05 | 29.9 | |
| BLIP-2Params=1.2B, Evaluation Protocol=Zero-shot2024.05 | 29.4 | |
| BLIPParams=223M, Evaluation Protocol=Zero-shot2024.05 | 28.1 | |
| ALBEFEvaluation Protocol=Zero-shot2024.05 | 27.7 | |
| OFAEvaluation Protocol=Fine-tuned2024.05 | 27.2 | |
| UNITEREvaluation Protocol=Fine-tuned2024.05 | 25.7 | |
| ViBERTEvaluation Protocol=Fine-tuned2024.05 | 24.5 | |
| CLIPParams=473M, Evaluation Protocol=Zero-shot2024.05 | 22.4 | |
| UNITEREvaluation Protocol=Zero-shot2024.05 | 19.8 | |
| ViBERTEvaluation Protocol=Zero-shot2024.05 | 19.3 |