Design and Implementation of a Tilapia (Oreochromis niloticus) Fingerling Counter Using Otsu Thresholding and Morphological Transformation Techniques

Many different image processing techniques have been developed and implemented in counting systems for livestock and fisheries applications. These existing systems often require sophisticated hardware and computational resources, leading to an increase in the financial costs associated with their production. In this paper, an innovative Tilapia (Oreochromis niloticus) fingerling counting system is described. The system is comprised of low-cost computer vision-based hardware and software that integrates Otsu thresholding and morphological transformation techniques. These techniques derive contours from segmented regions generated by the formulated algorithms. It uses a Raspberry Pi 4B single-board computer platform in conjunction with a Raspberry Pi Camera Module version 1.3 to capture images. The performance of the manual process and the proposed method were evaluated by determining its mean squared error using a ground truth as a reference. Based on the tests performed, it was revealed that the proposed automatic counting system resulted in a better performance than the manual counting method.