A Holistic Marine Video Dataset
Qianqian Li, Jiahua Li, Zhensheng Shi, Zhaorui Gu, Haiyong Zheng, Bing Zheng, Juan Li · OCEANS 2021: San Diego – Porto · 2021
With the rapid development of deep learning, many algorithms based on deep learning have been able to effectively help people solve many practical problems. In the field of marine research, it is very necessary to solve practical problems with the help of deep learning. In order to truly simulate the complex marine environment, we integrate a large amount of data including marine scenes, marine organisms and their actions to produce an marine video dataset named Holistic Marine Video Dataset (HMV) with abundant semantic aspects for marine scientific research. What’s more, we also provide baseline experiments of three tasks (marine organisms detection, marine scenes recognition and marine organism actions recognition) on HMV for reference.