| Abstrakt: |
Insects present in stored grain, wood, soil, plants, and environments have distinctive sets of acoustic features. This paper developed an insect detection and classification system using their sound dataset. A novel approach has been proposed based on the combination of features: Mel-frequency cepstral coefficient and Hilbert Huang transform, named Mel-frequency Hilbert Huang Transform (MFHTT) for acoustic feature extraction. The proposed method integrates the ability of principal component analysis (PCA) to reduce the dimensions and de-correlate the coefficients for insect sound classification. Support vector machine (SVM), K-nearest neighbour (KNN), Random Forest, Naïve Bayes and neural network have been analysed for achieving the highest detection accuracy. Experimental results show that the proposed feature extraction method performed better than baseline features and achieved an improved accuracy of 23.19% with the classifier KNN. Also, KNN outperforms as compared to other classifiers with a detection accuracy of 98.8%. |
