Weight Prediction System for Nile Tilapia using Image Processing and Predictive Analysis

Abstract

Fish farmers are likely to cultivate poor quality fish to accommodate the rising demands for food due to the everincreasing population. Fish growth monitoring greatly helps on producing higher quality fish products which leads to a better impact in the aquatic animal food production industry. However, monitoring through manual weighing and measuring stresses them that affects their health resulting to poorer quality or even fish kills. This paper presents a low-cost monitoring and Hough gradient method-based weight prediction system for Nile Tilapia (Oreochromis niloticus) using Raspberry Pi microcontroller and two low-cost USB cameras. This study aims to improve fish growth rate through monitoring the growth of the fishes with image processing eliminating the traditional way of obtaining fish measurements. By using paired t-test, the acquired values imply that the weight algorithm used to measure the weight of the fishes is accurate and acceptable to use. Growth performance of 10 Nile Tilapia was obtained in two intensive aquaculture setups-one for automated fish weighing through image processing and predictive analysis and the other setup for manual weighing. In response to weight prediction application, the growth of the fishes increased by 47.88%.

Morphological traits for allometric scaling of the European Sea Bass Dicentrarchus labrax (Linnaeus, 1758) from Southern Portugal population

“…Despite the well‐known advantages of these accessible technological methodologies, in terms of high accuracy and reduced fish manipulation, some strategies are important for mitigating uncertainties, such as image distortion and/or errors due to different operators (Gutzmann et al ., 2022). To address these drawbacks, image processing using machine learning, neural networks and computer vision techniques may provide a wide variety of useful algorithms from image acquisition, noise removal, image sharpening, image smoothing, image blurring to image segmentation, object detection, object recognition, feature extraction and classification (Awalludin et al ., 2020; Balaban, Chombeau et al ., 2010; Navarro et al ., 2016; Tolentino et al ., 2020; Zhang et al ., 2020). Exploring these methods may present an avenue for ongoing and future research.…”

Non-Invasive Fish Biometrics for Enhancing Precision and Understanding of Aquaculture Farming through Statistical Morphology Analysis and Machine Learning

“…In [25], a low-cost stereo vision monitoring system with the Hugh gradient method and 3rd-degree polynomial-based weight estimation capabilities for Nile tilapia (Oreochromis niloticus) using Raspberry Pi and low-cost USB cameras is proposed. It extracts the contour and then follows a pixel/metric conversion to estimate length.…”

Inferring Broiler Chicken Weight through Machine Learning

“…Tolentino et al [Tolentino et al 2020] realizou inferência de peso para tilápias do Nilo em ambiente aquático. Ao contrário de Amreai e Dohomen, cujos animais estão a uma distância fixa do dispositivo de gravac ¸ão, os peixes podem estar a diferentes distâncias, conforme se deslocam no aquário, afastando-se ou aproximando-se da câmera.…”