Runwei Guan, Ningwei Ouyang, Tianhao Xu, Shaofeng Liang, Wei Dai, Yafeng Sun, Shang Gao, Songning Lai, Shanliang Yao, Xuming Hu, Ryan Wen Liu, Yutao Yue, Hui Xiong
The paper introduces WaterCaption, a dataset for waterway image captioning, and Da Yu, a model that excels in generating detailed captions for waterway environments.
This research focuses on improving how unmanned surface vessels (USVs) understand their surroundings by using image captioning, which involves generating descriptive text from images. The authors created a new dataset called WaterCaption, specifically for waterways, to help train models to better describe these environments. They also developed a model named Da Yu, which uses a novel technique to efficiently generate detailed captions from images. This work aims to enhance the ability of USVs to monitor and understand complex waterway scenes more effectively.
