Fish monitoring and sizing using computer vision

Abstract

This paper proposes an image processing algorithm, based in a non invasive 3D optical stereo system and the use of computer vision techniques, to study fish in fish tanks or pools. The proposed technique will allow to study biological variables of different fish species in underwater environments. This knowledge, may be used to replace traditional techniques such as direct observation, which are impractical or affect the fish behavior, in task such as aquarium and fish farm management or fishway evaluation. The accuracy and performance of the proposed technique has been tested, conducting different assays with living fishes, where promising results were obtained.