| dc.description.abstract | Stereo image analysis of free-swimming farmed Atlantic salmon is today used for purposes such as individual size estimation and sea lice counting. This technology may in the future be used to score welfare and life history traits. This thesis aims for answering: 1. whether current stereo camera image analysis of size estimation reflects the true size distribution of caged salmon, and 2. whether morphometrical relationships of individual salmon, measured in images, can provide novel insights to growth performance and detection of sexual maturation. The data was collected from an experimental cage production with individually PIT-tagged salmon (n=4500 and n=2786 at harvest) that were manually size recorded multiple times over the production cycle. Stereo images were taken within the sea cage during the last 6 months prior to harvest. Individual images taken at harvest were linked with PIT-ID to enable individual comparison of recorded morphometrics with growth history. Stereo images allow for frequent and numerous measurements of free-swimming salmon, but the precision in estimating individual fish size and accuracy of size distribution within sea cages are largely undocumented. Weight estimations of free-swimming fish based on stereo image analysis are here compared with the true size distribution of the fish at the average weights 2.0, 4.0 and 5.5kg, including a comparison at an individual level by the stereo images taken at harvest. The results show that stereo image analysis gives a highly accurate weight estimation on an individual and populational level, but can to some degree be sensitive to fish size segregation in swimming depth. Morphometrical relationships of the salmon body, and knowledge of how this reflects the growth performance and sexual maturation is largely underexplored. The morphometric analysis includes the ratios of body height central/anal, eye diameter and head size vs. standard length, and are here compared with the harvest size of the fish, individual growth, and sexual maturation status. Harvest weight was reflected in growth rate already from ~50g size, and the largest fish had the highest body height central ratio, and the smallest fish had the largest head ratio. Sexually mature fish showed a clear difference to immature salmon for body height anal and head size, which may be used for detecting early signs of maturation. In conclusion, this thesis proves current stereo image analysis of size estimation as useful, and validate novel salmon morphometrics as relevant parameters for automatization in stereo image analysis. | |
