Video Question Answering on TVQA (test-public)
70.23AccuracySTAGE backbone
Evaluation Results
| Method | Links | |
|---|---|---|
| STAGE backboneTemporal Supervision=true2019.04 | 70.23 | |
| STAGE backboneTemporal Supervision=false2019.04 | 69.67 | |
| TVQA (S+V+Q)Timestamp Annotation=w/ ts, Video Feature=Visual Concept (cpt)2018.09 | 68.48 | |
| TVQA (S+V+Q)Timestamp Annotation=w/ ts, Video Feature=Regional (reg)2018.09 | 68.19 | |
| multi-taskTemporal Supervision=true2019.04 | 67.05 | |
| TVQA (S+V+Q)Timestamp Annotation=w/ ts, Video Feature=ImageNet (img)2018.09 | 66.94 | |
| PAMNTemporal Supervision=false2019.04 | 66.77 | |
| TVQA (S+V+Q)Timestamp Annotation=w/o ts, Video Feature=Visual Concept (cpt)2018.09 | 66.46 | |
| two-streamTemporal Supervision=false2019.04 | 66.46 | |
| TVQA (S+Q)Timestamp Annotation=w/ ts2018.09 | 66.36 | |
| TVQA (S+V+Q)Timestamp Annotation=w/o ts, Video Feature=ImageNet (img)2018.09 | 63.44 | |
| TVQA (S+V+Q)Timestamp Annotation=w/o ts, Video Feature=Regional (reg)2018.09 | 63.06 | |
| TVQA (S+Q)Timestamp Annotation=w/o ts2018.09 | 62.69 | |
| NNS-TFIDF STimestamp Annotation=w/ ts2018.09 | 50.79 | |
| NNS-TFIDF STimestamp Annotation=w/o ts2018.09 | 49.59 | |
| TVQA (V+Q)Timestamp Annotation=w/ ts, Video Feature=Visual Concept (cpt)2018.09 | 45.44 | |
| TVQA (V+Q)Timestamp Annotation=w/ ts, Video Feature=Regional (reg)2018.09 | 45.23 | |
| TVQA (V+Q)Timestamp Annotation=w/ ts, Video Feature=ImageNet (img)2018.09 | 43.91 | |
| TVQA (V+Q)Timestamp Annotation=w/o ts, Video Feature=Visual Concept (cpt)2018.09 | 43.84 | |
| TVQA (Q only)Timestamp Annotation=w/o ts2018.09 | 43.5 | |
| TVQA (Q only)Timestamp Annotation=w/ ts2018.09 | 43.5 | |
| TVQA (V+Q)Timestamp Annotation=w/o ts, Video Feature=Regional (reg)2018.09 | 43.01 | |
| TVQA (V+Q)Timestamp Annotation=w/o ts, Video Feature=ImageNet (img)2018.09 | 42.87 | |
| NNS-SkipThought STimestamp Annotation=w/ ts2018.09 | 38.29 | |
| Longest AnswerTimestamp Annotation=w/o ts2018.09 | 30.22 | |
| Longest AnswerTimestamp Annotation=w/ ts2018.09 | 30.22 | |
| NNS-Glove STimestamp Annotation=w/ ts2018.09 | 29.95 | |
| NNS-SkipThought STimestamp Annotation=w/o ts2018.09 | 26.93 | |
| Retrieval-SkipThoughtTimestamp Annotation=w/o ts2018.09 | 24.27 | |
| Retrieval-SkipThoughtTimestamp Annotation=w/ ts2018.09 | 24.27 | |
| NNS-Glove STimestamp Annotation=w/o ts2018.09 | 23.74 | |
| NNS-SkipThought QTimestamp Annotation=w/o ts2018.09 | 23.39 | |
| NNS-SkipThought QTimestamp Annotation=w/ ts2018.09 | 23.39 | |
| Retrieval-GloveTimestamp Annotation=w/o ts2018.09 | 22.77 | |
| Retrieval-GloveTimestamp Annotation=w/ ts2018.09 | 22.77 | |
| NNS-Glove QTimestamp Annotation=w/o ts2018.09 | 22.63 | |
| NNS-Glove QTimestamp Annotation=w/ ts2018.09 | 22.63 | |
| Retrieval-TFIDFTimestamp Annotation=w/o ts2018.09 | 20.78 | |
| Retrieval-TFIDFTimestamp Annotation=w/ ts2018.09 | 20.78 | |
| RandomTimestamp Annotation=w/o ts2018.09 | 20 | |
| RandomTimestamp Annotation=w/ ts2018.09 | 20 | |
| NNS-TFIDF QTimestamp Annotation=w/o ts2018.09 | 19.47 | |
| NNS-TFIDF QTimestamp Annotation=w/ ts2018.09 | 19.47 |