For our project, we are used the ImageNet dataset. The images are organized according to the WordNet, meaning related words and word phrases are grouped into one meaningful concept called a “synonym set” (aka “synset”). The dataset contains over 100k distinct synsets, with over 80k of them being nouns. Each synset contains an average of 1k images. This dataset is human annotated and is quality controlled. The average image size is 469×387.