1 paper
Qing Zhou, Junlin Huang, Qiang Li +2
From paired image-text training to text-only training for image captioning, the pursuit of relaxing the requirements for high-cost and large-scale annotation of good quality data r…