The deep seafloor is closest to the Earth's interior and there are complicated interactions in the physical and biological processes of the seafloor. Oceanographic parameters such as temperature, salinity, and ocean current can provide support for climate prediction. Therefore, the long-term observation of the deep seafloor is critical to the study of global climate change and the biodiversity in complex environments. For the requirements of long-term observation and remote data transmission in the deep sea of the South China Sea, a seafloor observation prototype system is designed in this paper as a stand-alone observation platform. We then proposed a date management method based on Controller Area Network to ensure the data quality. Distributed data management can be performed through system layering, and strict data transmission standards are established between each layer. At the same time, aiming at the difficulty of seafloor video data transmission in real-time, a new solution based on satellite communication technology, object detection technology, and high-efficiency compression coding technology of video images is proposed to provide support for seafloor biodiversity research. The observation data from the experiment are given, and these results show that the system can meet the requirements of long-term observation and remote real-time data transmission, and the reliability of the system is verified.