"EN-CLDI" (en-cldi) contains 1690 classes, which contains images paired with verbs and adjectives. Each word within this set is unique and paired with at least 22 images. It is the English subset of CLDI (cross-lingual dictionary induction) dataset from (Hartmann and Søgaard, 2018).

from datasets import load_dataset

common_words = load_dataset("jaagli/en-cldi", split="train")