Captioning Images with Diverse Objects

[J]. computer vision and pattern recognition, 2017: 1170-1178.