Dense Captioning on Visual Genome 5,000 images (test)
0.264METEOR (EB)FCLN on EB
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| FCLN on EBNumber of proposals=300, Image resolution=720x600, Hardware=Titan X GPU2015.11 | 0.264 | 0.296 | 0.293 | 4.88 | 3.21 | 26.84 | 210 | 140 | 10 | 360 | |
| Our model (FCLN)Number of proposals=300, Image resolution=720x600, Hardware=Titan X GPU2015.11 | 0.264 | 0.273 | 0.305 | 5.24 | 5.39 | 27.03 | 90 | 140 | 10 | 240 | |
| Region RNNNumber of proposals=300, Image resolution=720x600, Hardware=Titan X GPU2015.11 | 0.221 | 0.244 | 0.272 | 1.07 | 4.26 | 21.9 | 210 | 2,950 | 10 | 3,170 | |
| Full image RNNNumber of proposals=300, Image resolution=720x600, Hardware=Titan X GPU2015.11 | 0.173 | 0.197 | 0.209 | 2.42 | 4.27 | 14.11 | 210 | 2,950 | 10 | 3,170 |