The aim of this paper is to present automated fish fingerlings counting using image processing technique and to investigate the effectiveness of Convolutional Neural Network (CNN) in fish detection and counting accuracy. The proposed technique was tested in four different sizes of tilapia fingerlings – size 14, 17, 22 and 32. Threshold value was set to increase the level of efficiency of fish detection and counting accuracy. Experimental method and capturing images of fingerlings were conducted in the Bureau of Fisheries and Aquatic Resources (BFAR) Tilapia Hatchery and Fish Health Laboratory in Barcenaga, Naujan, Oriental Mindoro, Philippines. 2400 images of tilapia fingerlings were used in training and 1600 images in testing stage. The results of the experiment showed that CNN can be used in hatchery production and obtained a high level of fish detection and counting accuracy.
Paper
Full text
Automated Fingerlings Counting Using Convolutional Neural Network
Semantic Scholar · Computer Science · 2019
Abstract
The aim of this paper is to present automated fish fingerlings counting using image processing technique and to investigate the effectiveness of Convolutional Neural Network (CNN) in fish detection and counting accuracy. The proposed technique was tested in four different sizes of tilapia fingerlings – size 14, 17, 22 and 32. Threshold value was set to increase the level of efficiency of fish detection and counting accuracy. Experimental method and capturing images of fingerlings were conducted in the Bureau of Fisheries and Aquatic Resources (BFAR) Tilapia Hatchery and Fish Health Laboratory in Barcenaga, Naujan, Oriental Mindoro, Philippines. 2400 images of tilapia fingerlings were used in training and 1600 images in testing stage. The results of the experiment showed that CNN can be used in hatchery production and obtained a high level of fish detection and counting accuracy.