Underwater Real-Time Fish Recognition by Image Processing

Mehdi Chouiten


Pattern recognition, Image processing, FABEMD


Oceans and seas cover more than 70% of earth’s surface. However, only a very small proportion of people know the underwater world. The challenging objective of this work is to help people to identify fish species. This can be performed in several situations. It is destined to in-vivo divers either they are professional or not, snorkelers, aquarium visitors and any fish recognition task. The applications can be for wildlife monitoring and for leisure purposes. Our approach is to enable such an application on a smartphone protected by a waterproof diving case. Captured images are processed through an algorithm allowing classifying detected fishes in the correct specie. This classification is done using several algorithms based on different criteria detailed in the article.

Important Links:

Go Back