In the understanding of ecosystem dynamics, foraging behavior plays an important role. The availability of tracking data and the development of powerful statistical and computational tools have greatly enhanced foraging movement analysis. Estimation of behavioral modes within foraging trips has been classically done using state-space models, thus coupling a statistical model of variables obtained from tracking data, with a model of behavioral modes where the future mode depends on the preceding ones. Among the state-space models, Hidden Markov models have been preferred. Due to the lack of data for which the true behavioral modes are known, no real independent validation of the models have been previously done. The only predator for whom we can have access to true foraging behavioral modes in a natural environment is the human being. Here, using simultaneous tracking and ground-truthed data collected on fishers in the Humboldt Current System, we are able to do reliable and independent validation on our models. We compare the performance of different Markovian (Hidden Markov and semi-Markov) as well as discriminative models. Results support the use of Markovian models and particularly the semi-Markov model. Further perspectives regarding methods and ecological clues for pattern recognition in foraging movement are presented.