Visual Question Answering on VQA-CP v2 (test)
77.23Overall AccuracyTPCLDyn↑
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TPCLDyn↑Base=LXMERT, Year=2024, CL=Main technique2024.11 | 77.23 | — | — | — | — | |
| TPCLFix↑Base=LXMERT, Year=2024, CL=Main technique2024.11 | 75.83 | — | — | — | — | |
| GenBBase=LXMERT, Year=2023, Data Augmentation=Main technique2024.11 | 71.16 | — | — | — | — | |
| D-VQABase=LXMERT, Year=2021, Ensemble Learning=Main technique2024.11 | 69.75 | — | — | — | — | |
| SIMPLEAUGBase=LXMERT, Year=2021, Data Augmentation=Main technique2024.11 | 62.24 | — | — | — | — | |
| D-VQABase=UpDn2022.08 | 61.91 | 88.93 | 52.32 | 50.39 | — | |
| MUTANTBase=UpDn2022.08 | 61.72 | 88.9 | 49.68 | 50.78 | — | |
| MutantBase=UpDn, Year=2020, Data Augmentation=Main technique2024.11 | 61.72 | — | — | — | — | |
| DGGBase=UpDn, Year=2023, Data Augmentation=Main technique2024.11 | 61.14 | — | — | — | — | |
| FAN-VQABase=UpDn, Year=2024, Ensemble Learning=Secondary technique, Data Augmentation=Main technique2024.11 | 60.99 | — | — | — | — | |
| LBCLBase=UpDn, Year=2021, Ensemble Learning=Main technique, CL=Secondary technique2024.11 | 60.74 | — | — | — | — | |
| KDDAugBase=UpDn2022.08 | 60.24 | 86.13 | 55.08 | 48.08 | — | |
| CSS + IntroDBase=UpDn2022.08 | 60.17 | 89.17 | 46.91 | 48.62 | — | |
| GenBBase=UpDn2022.08 | 59.15 | 88.03 | 40.05 | 49.25 | — | |
| CSSBase=UpDn2022.08 | 58.95 | 84.37 | 49.42 | 48.21 | — | |
| LMH+CSSBase architecture=UpDown [3]2021.04 | 58.95 | — | — | — | — | |
| CSS*Base=LMH, extra_annotations=true2021.07 | 58.21 | 83.65 | 40.73 | 48.14 | 8.81 | |
| SSLBase=UpDn2022.08 | 57.59 | 86.53 | 29.87 | 50.03 | — | |
| GGEBase=UpDn2022.08 | 57.32 | 87.04 | 27.75 | 49.59 | — | |
| GGE-DQ-togBase=UpDn2021.07 | 57.32 | 87.04 | 27.75 | 49.59 | 15.27 | |
| GGE-DQ-iterBase=UpDn2021.07 | 57.12 | 87.35 | 26.16 | 49.77 | 16.44 | |
| RandImgBase=UpDn2022.08 | 55.37 | 83.89 | 41.6 | 44.2 | — | |
| RandImgBase architecture=UpDown [3]2021.04 | 55.37 | — | — | — | — | |
| CF-VQA(SUM) + IntroDBase=S-MRL2022.08 | 55.17 | 90.79 | 17.92 | 46.73 | — | |
| CF-VQA(SUM)Base=S-MRL2022.08 | 55.05 | 90.61 | 21.5 | 45.61 | — | |
| CF-VQA(Sum)Base=S-MRL2021.07 | 54.95 | 90.56 | 21.88 | 45.36 | — | |
| MFEBase=LMH2021.07 | 54.55 | 74.03 | 49.16 | 45.82 | — | |
| LMH + RMFEBase architecture=UpDown [3]2021.04 | 54.55 | — | — | — | — | |
| CF-VQA(Sum)Base=UpDn2021.07 | 53.69 | 91.25 | 12.8 | 45.23 | — | |
| CF-VQA(SUM)Base=UpDn2022.08 | 53.55 | 91.15 | 13.03 | 44.97 | — | |
| CF-VQABase=UpDn, Year=2021, Ensemble Learning=Main technique2024.11 | 53.55 | — | — | — | — | |
| LMHBase=UpDn2021.07 | 52.73 | 72.95 | 31.9 | 47.79 | 10.6 | |
| LMHBase=UpDn2022.08 | 52.45 | 69.81 | 44.46 | 45.54 | — | |
| Learned Mixin +H2019.09 | 52.05 | — | — | — | — | |
| LMHBase architecture=UpDown [3]2021.04 | 52.05 | — | — | — | — | |
| LMHBase=UpDn, Year=2019, Ensemble Learning=Main technique2024.11 | 52.01 | — | — | — | — | |
| VGQEBase=S-MRL2022.08 | 50.11 | 66.35 | 27.08 | 46.77 | — | |
| GVQE*Base=S-MRL, extra_annotations=true2021.07 | 50.11 | 66.35 | 27.08 | 46.77 | — | |
| UpDn+SCRExpl.=VQA-X2019.05 | 49.45 | 72.36 | 10.93 | 48.02 | — | |
| SCRBase=UpDn2022.08 | 49.45 | 72.36 | 10.93 | 48.02 | — | |
| SCR*Base=UpDn, extra_annotations=true2021.07 | 49.45 | 72.36 | 10.93 | 48.02 | — | |
| UpDn+SCRExpl.=HAT2019.05 | 49.17 | 71.55 | 10.72 | 47.49 | — | |
| DLRBase=UpDn2022.08 | 48.87 | 70.99 | 18.72 | 45.57 | — | |
| DLPBase=UpDn2021.07 | 48.87 | 70.99 | 18.72 | 45.57 | — | |
| DLRBase=UpDn, Year=2020, Ensemble Learning=Main technique2024.11 | 48.87 | — | — | — | — | |
| LMBase=UpDn2021.07 | 48.78 | 70.37 | 14.24 | 46.42 | 11.33 | |
| VGQEBase=UpDn2022.08 | 48.75 | — | — | — | — | |
| GVQE*Base=UpDn, extra_annotations=true2021.07 | 48.75 | — | — | — | — | |
| Learned Mixin2019.09 | 48.69 | — | — | — | — | |
| LXMERTBase=-, Year=20192024.11 | 48.66 | — | — | — | — | |
| ESRBase architecture=UpDown [3]2021.04 | 48.5 | — | — | — | — | |
| UpDn+SCRExpl.=QA2019.05 | 48.47 | 70.41 | 10.42 | 47.29 | — | |
| SCRBase=UpDn, Year=2019, Answer Re-Ranking=Main technique2024.11 | 48.47 | — | — | — | — | |
| UpDn+HINTExpl.=HAT2019.05 | 47.7 | 70 | 10.7 | 46.3 | — | |
| HINT*Base=UpDn, extra_annotations=true2021.07 | 47.5 | 67.21 | 10.67 | 46.8 | 10.34 | |
| RUBiBackbone=Baseline architecture (ours), Visual features=Bottom-Up Top-Down [15]2019.06 | 47.11 | 68.65 | 20.28 | 43.18 | — | |
| RUBiBase=S-MRL2021.07 | 47.11 | 68.65 | 20.28 | 43.18 | — | |
| HINTBase=UpDn2022.08 | 46.73 | 67.27 | 10.61 | 45.88 | — | |
| HINTBase=UpDn, Year=2019, Answer Re-Ranking=Main technique2024.11 | 46.73 | — | — | — | — | |
| NSM2019.07 | 45.8 | — | — | — | — | |
| RUBiBase=UpDn2021.07 | 45.42 | 63.03 | 11.91 | 44.33 | 6.27 | |
| RUBiBase=UpDn2022.08 | 44.23 | 67.05 | 17.48 | 39.61 | — | |
| RUBiBase architecture=UpDown [3]2021.04 | 44.23 | — | — | — | — | |
| RUBiBase=UpDn, Year=2019, Ensemble Learning=Main technique2024.11 | 44.23 | — | — | — | — | |
| Grand and Belinkov2019.09 | 42.33 | — | — | — | — | |
| CVLBase=UpDn2022.08 | 42.12 | 45.72 | 12.45 | 48.34 | — | |
| Q-type Balanced SamplingBackbone=Baseline architecture (ours), Visual features=Bottom-Up Top-Down [15]2019.06 | 42.11 | 61.55 | 11.26 | 40.39 | — | |
| UpDn+AdvReg.2019.05 | 41.2 | 65.5 | 15.5 | 35.5 | — | |
| Ramakrishnan et al.2019.09 | 41.17 | — | — | — | — | |
| UpDn + Q-Adv + DoEVisual features=Bottom-Up Top-Down [15]2019.06 | 41.17 | 65.49 | 15.48 | 35.48 | — | |
| ARegBase=UpDn2022.08 | 41.17 | 65.49 | 15.48 | 35.48 | — | |
| AdvReg.Base=UpDn2021.07 | 41.17 | 65.49 | 15.48 | 35.48 | — | |
| CSS*Base=UpDn, extra_annotations=true2021.07 | 41.16 | 43.96 | 12.78 | 47.48 | 8.23 | |
| REGAT2019.07 | 40.42 | — | — | — | — | |
| Balanced SamplingBackbone=Baseline architecture (ours), Visual features=Bottom-Up Top-Down [15]2019.06 | 40.38 | 57.99 | 10.07 | 39.23 | — | |
| Reweight2019.09 | 40.06 | — | — | — | — | |
| UpDn2022.08 | 39.94 | 42.46 | 11.93 | 45.09 | — | |
| Bias Product2019.09 | 39.93 | — | — | — | — | |
| UpDnBase=N/A2021.07 | 39.89 | 43.01 | 12.07 | 45.82 | 3.91 | |
| UpDnTrained by=[25], Visual features=Bottom-Up Top-Down [15]2019.06 | 39.74 | 42.27 | 11.93 | 46.05 | — | |
| UpDown2021.04 | 39.74 | — | — | — | — | |
| UpDnBase=-, Year=20182024.11 | 39.74 | — | — | — | — | |
| UpDn2019.05 | 39.7 | 42.7 | 11.9 | 46.1 | — | |
| MuRelBottom-up features=true2019.02 | 39.54 | 42.85 | 13.17 | 45.04 | — | |
| MuRel2019.07 | 39.54 | — | — | — | — | |
| MuRelVisual features=Bottom-Up Top-Down [15]2019.06 | 39.54 | 42.85 | 13.17 | 45.04 | — | |
| LfFBase architecture=UpDown [3]2021.04 | 39.49 | — | — | — | — | |
| AttAlignBase=UpDn, Year=2019, Answer Re-Ranking=Main technique2024.11 | 39.37 | — | — | — | — | |
| BAN2019.07 | 39.31 | — | — | — | — | |
| BANTrained by=[45], Visual features=Bottom-Up Top-Down [15]2019.06 | 39.31 | — | — | — | — | |
| RAMEN2019.07 | 39.21 | — | — | — | — | |
| RAMENVisual features=Bottom-Up Top-Down [15]2019.06 | 39.21 | — | — | — | — | |
| None2019.09 | 39.18 | — | — | — | — | |
| BLOCK2021.04 | 38.69 | — | — | — | — | |
| UpDn+AttAlign2019.05 | 38.5 | 42.5 | 11.4 | 43.8 | — | |
| Baseline architectureVisual features=Bottom-Up Top-Down [15]2019.06 | 38.46 | 42.85 | 12.81 | 43.2 | — | |
| S-MRL2022.08 | 38.46 | 42.85 | 12.81 | 43.2 | — | |
| S-MRLBase=N/A2021.07 | 38.46 | 42.85 | 12.81 | 43.2 | — | |
| AttentionBottom-up features=true2019.02 | 38.04 | 41.56 | 12.19 | 43.29 | — | |
| UpDnTrained by=[45], Visual features=Bottom-Up Top-Down [15]2019.06 | 38.01 | — | — | — | — |