Visual Dialog on VisDial 0.5 (test)
0.635MRRHuman Performance
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Human PerformanceParadigm=Human, Input Context=Question, Image, History2016.11 | 0.635 | 48.03 | 83.76 | — | 2.83 | |
| Human PerformanceParadigm=Human, Input Context=Question, Image2016.11 | 0.619 | 46.12 | 82.54 | — | 2.92 | |
| Memory Network (MN)Paradigm=Discriminative, Input Context=Question, Image, History2016.11 | 0.529 | 37.33 | 68.47 | 79.54 | 7.03 | |
| Memory Network (MN)Paradigm=Discriminative, Input Context=Question, History2016.11 | 0.524 | 36.84 | 67.78 | 78.92 | 7.25 | |
| Late Fusion (LF)Paradigm=Discriminative, Input Context=Question, Image, History2016.11 | 0.511 | 36.72 | 67.46 | 78.3 | 7.63 | |
| HieCoAttParadigm=VQA / Discriminative, Input Context=Question, Image2016.11 | 0.509 | 35.54 | 66.79 | 77.94 | 7.68 | |
| HREAParadigm=Discriminative, Input Context=Question, Image, History2016.11 | 0.508 | 36.76 | 66.54 | 77.75 | 7.59 | |
| SANParadigm=VQA / Discriminative, Input Context=Question, Image2016.11 | 0.506 | 36.21 | 67.08 | 78.16 | 7.74 | |
| Late Fusion (LF)Paradigm=Discriminative, Input Context=Question, History2016.11 | 0.505 | 36.21 | 66.56 | 77.31 | 7.89 | |
| Late Fusion (LF)Paradigm=Discriminative, Input Context=Question, Image2016.11 | 0.502 | 35.76 | 66.59 | 77.61 | 7.72 | |
| Hierarchical Recurrent Encoder (HRE)Paradigm=Discriminative, Input Context=Question, Image, History2016.11 | 0.502 | 36.26 | 65.67 | 77.05 | 7.79 | |
| Hierarchical Recurrent Encoder (HRE)Paradigm=Discriminative, Input Context=Question, History2016.11 | 0.489 | 34.74 | 64.25 | 75.4 | 8.32 | |
| Human PerformanceParadigm=Human, Input Context=Question, History2016.11 | 0.485 | 30.31 | 70.53 | — | 3.91 | |
| Late Fusion (LF)Paradigm=Discriminative, Input Context=Question2016.11 | 0.482 | 34.29 | 63.42 | 74.31 | 8.87 | |
| Memory Network (MN)Paradigm=Generative, Input Context=Question, Image, History2016.11 | 0.443 | 34.62 | 53.74 | 60.18 | 21.69 | |
| Hierarchical Recurrent Encoder (HRE)Paradigm=Generative, Input Context=Question, Image, History2016.11 | 0.442 | 34.37 | 53.4 | 59.74 | 21.75 | |
| HREAParadigm=Generative, Input Context=Question, Image, History2016.11 | 0.442 | 34.47 | 53.43 | 59.73 | 21.83 | |
| Human PerformanceParadigm=Human, Input Context=Question2016.11 | 0.441 | 25.1 | 67.37 | — | 4.19 | |
| Late Fusion (LF)Paradigm=Generative, Input Context=Question, Image2016.11 | 0.437 | 34.06 | 52.5 | 58.89 | 22.31 | |
| Memory Network (MN)Paradigm=Generative, Input Context=Question, History2016.11 | 0.434 | 33.12 | 53.14 | 59.61 | 22.14 | |
| Late Fusion (LF)Paradigm=Generative, Input Context=Question, Image, History2016.11 | 0.43 | 33.27 | 51.96 | 58.09 | 23.04 | |
| Hierarchical Recurrent Encoder (HRE)Paradigm=Generative, Input Context=Question, History2016.11 | 0.43 | 32.84 | 52.36 | 58.64 | 22.59 | |
| Late Fusion (LF)Paradigm=Generative, Input Context=Question, History2016.11 | 0.425 | 32.49 | 51.56 | 57.8 | 23.11 | |
| Late Fusion (LF)Paradigm=Generative, Input Context=Question2016.11 | 0.403 | 29.74 | 50.1 | 56.32 | 24.06 | |
| NN-QParadigm=Baseline, Input Context=Question2016.11 | 0.392 | 30.54 | 46.99 | 49.98 | 30.88 | |
| NN-QIParadigm=Baseline, Input Context=Question, Image2016.11 | 0.385 | 29.71 | 46.57 | 49.86 | 30.9 | |
| Answer priorParadigm=Baseline2016.11 | 0.311 | 19.85 | 39.14 | 44.28 | 31.56 |