Fish identification

This study explores the feasibility of individual fish identification based on the unique biometric patterns of Atlantic salmon appearance such as the body shape, position of eye, gills, fins or colour patterns on the skin or iris pattern determined by machine vision systems. The uniqueness of the visible patterns for the group of individual fish will be studied. Moreover, this study will investigate the long-term stability and possible changes of investigated patterns due to aging. The output of this study would enable individualized fish treatment which will lead not only to the optimization of feeding, real-time disease detection and precise biomass estimation but also it will be suitable substitute to invasive tagging methodology. Furthermore, it will enable fish farmers to implement The stress to individual fish is small, as the only stress the fish will experience is during netting out of the tank and during PIT-tagging and anesthesia. This procedure will use 300 experimental Atlantic salmon in three size classes ranging from 50 – 600 gram. For the long-term experiment, 30 fish will be tagged using PIT tagged before the data collection. The Experiment need to use Fish because the patterns is going to be determined from the fish skin. Also, the aging of the fish skin pattern must be studied on the real fish. The subset of 300 fish is the minimal number to be able to make a conclusion about usage of the fish visible patterns for identification. The anaesthesia and PIT-tagging of the fish, which is needed for the image data collection will be done by the trained personnel following the standard procedures.