Analysis of feature based object mining and tagging algorithm considering different levels of occlusion

• Published in: 2017 International Conference on Communication and Signal Processing (ICCSP) pp. 0171 - 0175
• Main Authors: Lad, Hiteshree H., Mehta, Mayuri A.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-04-2017
• Subjects: Algorithm design and analysis; Classification algorithms; Feature extraction; feature vector; object mining; Object recognition; object tagging; Support vector machines; Tagging; Training
• DOI: 10.1109/ICCSP.2017.8286800

Accurate identification and tagging of objects has been significant in several domains of computer science because applications in these domains use tags to recognize and to distinguish among the objects. We have proposed a novel Feature based Object Mining and Tagging Algorithm (FOMTA) that identifies the objects with higher accuracy from an occluded image. In this paper, we present the analysis of FOMTA considering images from three benchmark datasets. The experiments are performed considering different levels of occlusion as well as considering different sizes of an image. The experimental results show that the FOMTA outperforms the conventional region based approach. Specifically, it increases the object identification accuracy and thus, decreases the misclassification rate under different levels of occlusion. Moreover, it decreases the overall computational time.