MRRC: multiple role representation crossover interpretation for image captioning with R-CNN feature distribution composition (FDC).评价结果

评估详情