基于多视图跨模态特征融合的图像描述生成

张乃洲,赵云超,曹薇,张啸剑

Image captioning generation based on multiple-view cross-modal feature fusion

Naizhou ZHANG,Yunchao ZHAO,Wei CAO,Xiaojian ZHANG

表 3 在Flickr30k数据集上与其他先进模型的性能比较

Tab.3 Comparison with other state-of-the-art model on Flickr30k dataset

% 模型 BLEU-1 BLEU-4 METEOR ROUGE-L CIDEr Soft-Attention[4] 66.7 19.1 18.5 — — Hard-Attention[4] 66.9 19.9 18.5 — — Adaptive-Attention[5] 67.7 25.1 20.4 — 53.1 A_R_L[35] 69.8 27.7 21.5 48.5 57.4 IVAIC[36] 70.8 30.6 22.5 49.8 63.0 VRCDA[33] 73.2 30.6 22.7 50.6 66.0 MVCMFAF (本文) 75.2 33.7 34.2 52.1 75.6