Sentence Retrieval on Flickr30K
3,740R@1RTP (weighted distance)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| RTP (weighted distance)2015.04 | 3,740 | 6,310 | 7,430 | — | |
| RTP (SAE)2015.04 | 3,670 | 6,190 | 7,360 | — | |
| m-RNN-vggBackbone=VGG2015.04 | 3,540 | 6,380 | 7,370 | 3 | |
| FV (GMM+HGLMM)2015.04 | 3,500 | 6,200 | 7,380 | 3 | |
| FV (HGLMM)2015.04 | 3,440 | 6,100 | 7,230 | 3 | |
| m-CNNENSBackbone=VGG2015.04 | 3,360 | 6,410 | 7,490 | 3 | |
| FV (GMM)2015.04 | 3,300 | 6,070 | 7,190 | 3 | |
| FV (LMM)2015.04 | 3,250 | 5,990 | 7,150 | 3.2 | |
| RTP (joint training)2015.04 | 3,100 | 5,860 | 6,790 | — | |
| m-CNNstBackbone=VGG2015.04 | 2,700 | 5,640 | 7,010 | 4 | |
| m-CNNphsBackbone=VGG2015.04 | 2,500 | 5,480 | 6,680 | 4.5 | |
| FV (Mean Vec)2015.04 | 2,480 | 5,250 | 6,430 | 5 | |
| m-CNNphlBackbone=VGG2015.04 | 2,390 | 5,420 | 6,600 | 5 | |
| MNLM-vggBackbone=VGG2015.04 | 2,300 | 5,070 | 6,290 | 5 | |
| DVSA (BRNN)2015.04 | 2,220 | 4,820 | 6,140 | 4.8 | |
| m-CNNwdBackbone=VGG2015.04 | 2,130 | 5,320 | 6,610 | 5 | |
| m-CNNENSBackbone=OverFeat2015.04 | 2,010 | 4,420 | 5,630 | 8 | |
| DVSA (DepTree)2015.04 | 2,000 | 4,660 | 5,940 | 5.4 | |
| m-RNN2015.04 | 1,840 | 4,020 | 5,090 | 10 | |
| NIC2015.04 | 1,700 | — | 5,600 | 7 | |
| DCCA2015.04 | 1,670 | 3,930 | 5,290 | 8 | |
| MNLM2015.04 | 1,480 | 3,920 | 5,090 | 10 | |
| m-CNNstBackbone=OverFeat2015.04 | 1,480 | 3,790 | 4,980 | 11 | |
| m-CNNphsBackbone=OverFeat2015.04 | 1,440 | 3,860 | 4,960 | 11 | |
| Deep Fragment2015.04 | 1,420 | 3,770 | 5,130 | 10 | |
| m-CNNphlBackbone=OverFeat2015.04 | 1,380 | 3,810 | 4,850 | 11.5 | |
| m-CNNwdBackbone=OverFeat2015.04 | 1,270 | 3,020 | 4,450 | 14 | |
| RVP (T+I)2015.04 | 1,210 | 2,780 | 4,780 | 11 | |
| RVP (T)2015.04 | 1,190 | 2,500 | 4,770 | 12 | |
| SDT RNN2015.04 | 960 | 2,980 | 4,110 | 16 | |
| DeViSE2015.04 | 450 | 1,810 | 2,920 | 26 | |
| Random Ranking2015.04 | 10 | 60 | 110 | 631 |