Ferret-v2: An Improved Baseline for Referring and Grounding with Large Language Models
High-resolution visual grounding and referring through flexible image resolution, richer visual features, and staged training.

BibTeX citation
@inproceedings{ferretv22024,
title = {Ferret-v2: An Improved Baseline for Referring and Grounding with Large Language Models},
author = {Haotian Zhang and Haoxuan You and Philipp Dufter and Bowen Zhang and Chen Chen and Hong-You Chen and Tsu-Jui Fu and William Yang Wang and Shih-Fu Chang and Zhe Gan and Yinfei Yang},
booktitle = {COLM},
year = {2024},
url = {https://arxiv.org/abs/2404.07973}
}