docs/intro.md

DesCo

This is the HuggingFace Gradio Demo for the paper DesCo: Learning Object Recognition with Rich Language Descriptions (NeurIPS 2023).

ground_tokens specifies which tokens we wish the model to ground to, separated by ;; if it is not specified, the script will use NLTK to extract noun phrases automatically.

Code: https://github.com/liunian-harold-li/DesCo