Table 4 Experimental results on TRECVID AVS 2016, 2017, and 2018
From: Level-wise aligned dual networks for text–video retrieval
TRECVID edition | ||||
|---|---|---|---|---|
2016 | 2017 | 2018 | OVERALL | |
Top 3 TRECVID finalists | ||||
Rank 1 | 5.4 [53] | 20.6 [40] | 12.1 [54] | – |
Rank 2 | 5.1 [55] | 15.9 [39] | 8.7 [56] | – |
Rank 3 | 4 [57] | 12 [58] | 8.2 [59] | – |
Literature methods | ||||
8.7 | 15 | – | – | |
Markatopoulou et al. [37] | 6.4 | – | – | – |
CE [13] | 7.4 | 14.5 | 8.6 | 10.2 |
VSE++ [52] | 13.5 | 16.3 | 10.6 | 13.5 |
W2VV [42] | 14.9 | 19.8 | 10.3 | 15 |
W2VV++ [20] | 15.1 | 21.3 | 10.6 | 15.7 |
Dual Encoding [22] | 15.2 | 23.1 | 12.1 | 16.8 |
LADN | 15.3 | 24.1 | 12.6 | 17.3 |