Application and comparison of modified classifiers for human activity recognition

In this paper, custom modifications of Orthogonal Matching Pursuit and Self Organizing Maps based classification algorithms are used and compared to standard and widely used classification techniques with applications to human activity recognition. Seven algorithms are compared in terms of their accuracy performance. The modifications are described in this paper and shown to perform better than commonly used classifiers. The results indicate that human activities can be successfully and reliably recognized even without data preprocessing.

---

W artykule opisano klasyczne i rzadziej używane metody klasyfikacji danych używanych do rozpoznawania aktywności człowieka. Po równano szereg algorytmów oraz zmodyfikowano algorytm OMP w celu usunięcia ograniczeń.