Action Anticipation On Epic Kitchens 55 1
评估指标
Top 1 Accuracy - Act.
Top 1 Accuracy - Noun
Top 1 Accuracy - Verb
Top 5 Accuracy - Act.
Top 5 Accuracy - Noun
Top 5 Accuracy - Verb
评测结果
各个模型在此基准测试上的表现结果
模型名称 | Top 1 Accuracy - Act. | Top 1 Accuracy - Noun | Top 1 Accuracy - Verb | Top 5 Accuracy - Act. | Top 5 Accuracy - Noun | Top 5 Accuracy - Verb | Paper Title | Repository |
|---|---|---|---|---|---|---|---|---|
| ImagineRNN | 9.25 | 15.50 | 29.33 | 22.19 | 35.78 | 70.67 | Learning to Anticipate Egocentric Actions by Imagination | - |
| 2SCNN | 2.29 | 9.97 | 25.23 | 9.35 | 27.38 | 68.66 | Scaling Egocentric Vision: The EPIC-KITCHENS Dataset | |
| ATSN | 2.39 | 10.41 | 25.30 | 6.63 | 29.50 | 68.32 | Scaling Egocentric Vision: The EPIC-KITCHENS Dataset | |
| Abstract Goal | 13.28 | 22.36 | 41.41 | 24.24 | 41.62 | 73.10 | Predicting the Next Action by Modeling the Abstract Goal | - |
| RULSTM [24, 23] | 8.16 | 15.19 | 27.01 | 21.10 | 34.38 | 69.55 | What Would You Expect? Anticipating Egocentric Actions with Rolling-Unrolling LSTMs and Modality Attention | |
| ED | 2.65 | 7.81 | 22.52 | 7.57 | 21.42 | 62.65 | RED: Reinforced Encoder-Decoder Networks for Action Anticipation | |
| AVT+ | 10.41 | 15.64 | 30.66 | 24.27 | 40.76 | 72.17 | Anticipative Video Transformer |
0 of 7 row(s) selected.