Motion analysis of living animals might help to understand the dangers that threaten the species, but it is sometimes technically challenging. This is particularly true for biologists who study migrating fishes of the basin of Adour. Here, we investigate the use of computer vision for automatic tracking and acquisition of knowledge about the migratory behavior of glass eels. For that purpose, some glass eels are color-marked, then introduced into an experimental medium reproducing tidal conditions. Observations are collected in video sequences. To get information about fish behavior, one can track their motion by optical flow methods. Preliminary results show that one can estimate the velocity but also the swimming behavior, hence eventually leading to energetic information about the animals.