Abstract: Computer vision technology has emerged as a key technical means for the intelligent management of aquatic animal farming, thanks to its non-invasive, high-precision, and automated characteristics. This paper systematically reviews the applications of computer vision in two major aspects of aquatic animal farming: biomass statistics (including population counting, body length and weight estimation) and health status assessment (covering disease identification, deformity detection, and behavior monitoring). In terms of biomass statistics, the paper focuses on analyzing counting methods based on detection, regression, and tracking, as well as technologies for body length and weight estimation that rely on key points, contours, and stereo vision. For health status assessment, it delves into the application of body color, texture, and body shape features in disease and deformity identification, along with quantitative methods for behavior monitoring in the context of physiological activities, abnormal behaviors, and group dynamics. Furthermore, the paper points out the current challenges faced by the technology, such as difficulties in data acquisition, weak model generalization ability, strong environmental interference, and insufficient multi-modal fusion. It also puts forward prospects for future research directions, including the construction of cross-domain datasets, the development of adaptive models, and the integration of multiple technologies, aiming to promote the intelligent and precise development of aquaculture.