Comparative Study on Interaction of Form and Motion Processing Streams by Applying Two Different Classifiers in Mechanism for Recognition of Biological Movement
Figure 5
Confusion matrices SNN classifying KTH dataset obtained by adapted active basis model as combination of form and motion pathways. Confusion matrices of the proposed approach have been presented for the case without fuzzy interference system, left matrix, and, after it, right matrix which are achieved from human action movements of KTH dataset [39]. The robustness of the method after adding the fuzzy interference stabilizer is considerably increased. The wrong recognitions in the left confusion matrix have been decreased especially in case of some actions, that is, clapping. Moreover, soar of robustness helps increase the overall accuracy and gives better results in classification of biological movement. The accuracy of categorizations using unbalanced SNN is reached at 86.46%.
