The authors trained a linear mapping between the image representations and the word embeddings representing 8 classes for which they had labeled images, thus ...
確定! 回上一頁