Weight estimation model for red tilapia (Oreochromis niloticus Linn.) from images

Abstract

Importance of the work: The conventional method of weighing the fish is by hand, which is labor-intensive and can cause stress or injury to the fish. Objectives: To investigate a length-weight model of red tilapia and develop a computer program to estimate the weight from the length. Materials & Methods: Red tilapia were raised in three cages suspended in a pond. In total, 270 fish (weight 50-810 g) of the first crop were sampled every 2 wk and total length and weight were measured. A program was developed and validated using fish weights from the next crop. Results: The length-weight relationship was determined as y = 0.0196 x2.9868. The correlation coefficient was 0.9902, indicating a good correlation between the length and weight of the fish. The coefficient of determination was 0.9802, indicating a showing high-quality linear regression prediction. The growth coefficient, (b value of 2.9868) was not significantly (p > 0.05) different from the standard value of 3.0, indicating isometric growth, while the condition factor was in the range 0.81-1.98. These values were greater than 1, suggesting that the fish were in good condition. The program was developed and validated on 170 fish from the second crop. Accuracy in evaluating the body weight of the fish was 93.01 ± 3.11% compared to the conventional method. Main finding: The developed program showed acceptable weight accuracy and involved no requirement to catch the fish, thereby reducing fish stress and their possible injury, with time and labor reductions.