Aquaculture in modern conditions is one of the priority directions of human economic activity, aimed at rational use of aquatic biological resources and maintenance of ecological balance in freshwater and marine ecosystems, providing reproduction and cultivation of various hydrobionts, including fish, mollusks, crustaceans, and algae, under controlled conditions, allowing to regulate both the quality and quantity of raw materials. In recent years, given the significant number of water areas in Russia actively developing fish farms and biological laboratories for breeding and research of fish. Many tasks in fish production facilities that are performed by staff and are often time-consuming and resource-consuming are effectively handled with the use and development of digital technology. Tasks aimed at analyzing visual information with the help of artificial intelligence technologies and deep learning networks are presented in a wide range of directions. An automatic fish monitoring system was developed that uses the YOLOv9 neural network detector and StrongSORT tracking to measure the length, mass, and speed of each individual in real time and to construct full distributions of these parameters. In an experiment with 100 catfish fry, the system showed high accuracy (the measurement error does not exceed a few percent), while allowing the detection of extreme values and behavioral anomalies that are inaccessible with traditional manual methods.