The Identification of Hunger Behaviour of Lates Calcarifer through the Integration of Image Processing Technique and Support Vector Machine

4th Asia Pacific Conference on Manufacturing Systems; 3rd International Manufacturing Engineering Conference

Taha, Z.^1 ; Razman, M.A.M.^1 ; Adnan, F.A.^1 ; Abdul Ghani, A.S.^1 ; Abdul Majeed, A.P.P.^1 ; Musa, R.M.^1 ; Sallehudin, M.F.^2 ; Mukai, Y.^2

Innovative Manufacturing, Mechatronics and Sports Lab (IMAMS), Faculty of Manufacturing Engineering, Universiti Malaysia Pahang, Pekan Campus, Pekan, Pahang

26600, Malaysia^1

Kulliyyah of Science, International Islamic University Malaysia, Jalan Sultan Ahmad Shah, Bandar Indera Mahkota, Pahang, Kuantan

25200, Malaysia^2

关键词: Centre of gravity;    Classification accuracy;    Fish behaviours;    Integration of images;    Integration techniques;    Lates Clacarifer;    Machine learning techniques;    Reduction of oxygen;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/319/1/012028/pdf
DOI  :  10.1088/1757-899X/319/1/012028

来源: IOP

【 摘 要 】

Fish Hunger behaviour is one of the important element in determining the fish feeding routine, especially for farmed fishes. Inaccurate feeding routines (under-feeding or over-feeding) lead the fishes to die and thus, reduces the total production of fishes. The excessive food which is not eaten by fish will be dissolved in the water and thus, reduce the water quality (oxygen quantity in the water will be reduced). The reduction of oxygen (water quality) leads the fish to die and in some cases, may lead to fish diseases. This study correlates Barramundi fish-school behaviour with hunger condition through the hybrid data integration of image processing technique. The behaviour is clustered with respect to the position of the centre of gravity of the school of fish prior feeding, during feeding and after feeding. The clustered fish behaviour is then classified by means of a machine learning technique namely Support vector machine (SVM). It has been shown from the study that the Fine Gaussian variation of SVM is able to provide a reasonably accurate classification of fish feeding behaviour with a classification accuracy of 79.7%. The proposed integration technique may increase the usefulness of the captured data and thus better differentiates the various behaviour of farmed fishes.