Video Summarization on How2 1.0 (test)
3.89INFMulti-source Sequence-to-Sequence Model with Hierarchical Attention
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Multi-source Sequence-to-Sequence Model with Hierarchical AttentionModality=Text-and-Video, Model ID=8, Input Modalities=Ground-truth transcript + Action2019.06 | 3.89 | 3.74 | 3.85 | 3.94 | |
| S2S Complete TranscriptModality=Text-only, Model ID=5a, Input Context=650 tokens2019.06 | 3.86 | 3.78 | 3.78 | 3.92 | |
| Action Features + RNNModality=Video-only, Model ID=72019.06 | 3.58 | 3.3 | 3.71 | 3.8 |