| <p align="center" width="100%"> | |
| <img src="https://ice.frostsky.com/2024/03/17/214a3af4a34a26be0a04e551e16b9364.webp" width="40%" height="80%"> | |
| </p> | |
| # Chain-of-Spot: Interactive Reasoning Improves Large Vision-Language Models | |
| ### Model details: | |
| Chain-of-Spot encourages Large Vision-Language Models to identify the region of interest (ROI) in the image condition on the question and reasoning through an interactive manner, thereby improving the ability of visual understanding. | |
| ### Where to send questions or comments about the model: https://github.com/dongyh20/Chain-of-Spot | |
| ### Paper or resources for more information: https://sites.google.com/view/chain-of-spot/ |