Figure Captioning on SciCap First Sentence
15.53BLEUChaTS-Pi (DePLot+FLAN-T5)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ChaTS-Pi (DePLot+FLAN-T5)Input=DePLot table2024.05 | 15.53 | 0.48 | 0.3 | 0.18 | |
| DePLot+T5Input=image2024.05 | 15.27 | 0.3 | 0.28 | 0 | |
| ChaTS-Pi (DePLot+T5)Input=DePLot table2024.05 | 15.2 | 0.44 | 0.29 | 0.22 | |
| DePLot+FLAN-T5Input=image2024.05 | 15.12 | 0.34 | 0.29 | 0.18 | |
| MATCHA-DePLot+FLAN-T5Input=DePLot-table2024.05 | 14.89 | 0.36 | 0.29 | 0.2 | |
| ChaTS-Pi (MATCHA-DePLot+FLAN-T5)Input=image2024.05 | 14.19 | 0.49 | 0.3 | 0.49 | |
| PaLI-17BInput=image, resolution=5882024.05 | 11.05 | 0.41 | 0.3 | — | |
| CNN+LSTM(vision only)Input=image2024.05 | 2.19 | — | — | — | |
| PALM-2Input=image2024.05 | 0.94 | 0.84 | 0.3 | 0.36 | |
| ChaTS-Pi (PALM-2)Input=DePLot table2024.05 | 0.76 | 0.93 | 0.31 | 0.36 |