Image Captioning on MSCOCO 1,000 images 2014 (test)
24.6BLEU-4Google NIC
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Google NICBackbone=GoogLeNet2014.12 | 24.6 | 66.6 | 46.1 | 32.9 | — | — | |
| Multimodal RNNBackbone=VGGNet2014.12 | 23 | 62.5 | 45 | 32.1 | 19.5 | 66 | |
| MS Research2014.12 | 21.1 | — | — | — | 20.7 | — | |
| Chen and Zitnick2014.12 | 19 | — | — | — | 20.4 | — | |
| Nearest NeighborBackbone=VGGNet2014.12 | 10 | 48 | 28.1 | 16.6 | 15.7 | 38.3 | |
| LRCNBackbone=AlexNet2014.12 | — | 62.8 | 44.2 | 30.4 | — | — |