A Portable System for Automated Measurement of Striped Catfish Length Using Computer Vision

Abstract

In fish farming, regular measurement of fish length is a necessary work to evaluate the growth and health status of fish. However, manual measurement of fish length using measuring board is a time-consuming task and subject to error due to human bias factor. In this paper, the authors present the use of image processing techniques and smartphone cameras for automated measurement of striped catfish length. The system was designed to be portable by taking advantage of laptop computers and high-performance cameras found on most smartphones. The image processing algorithms include the following steps: take image of fish, convert image to binary one, find contours and draw masks, extract the region of interest, determine fish orientation, find the position of fish’s caudal peduncle, and calculate fish length. The system algorithms were implemented in Python programming language in combination with the OpenCV library. Experimental validation with 30 striped catfish samples shows that the designed system can provide an average accuracy of 97.71%. The proposed method could be a potential alternative of fish length measurement to support fisheries industry.