DEVELOPMENT OF ARTIFICIAL INTELLIGENCE SUPPORTED PUFFERFISH TRAP
Makaleler > Tam Makale · 2026
Yazarlar: Mert Demir
Konya Journal of Engineering Sciences, Uluslararası, 2026, cilt 14, ss 682-699
Özet
Due to changing climatic conditions, living things leave their habitats and migrate to other geographies. One of these creatures, the pufferfish, has created new habitats by migrating from warm seas to the Mediterranean and Aegean seas in recent years. The pufferfish is a dominant and invasive species, and its population is rapidly increasing wherever it is found because there is no other creature hunting it. Since it is a poisonous species, there is no possibility of biological control and hunting of this creature in fishing activities and controlling its population are encouraged, but this method is not effective. The scope of the study describes the development of a pufferfish trap developed with artificial intelligence models that targets the invasive pufferfish. Different sized trap designs have been suggested for different sized pufferfish species. To detect the pufferfish, 3200 pufferfish images and 3000 images of Mediterranean fish species were trained on the system with a supervised machine learning model to distinguish this species from other species. Addition, the calculation of trap size according to the species of this creature, which has different sizes, was handled with another artificial intelligence model. The developed model detects pufferfish from distances of 35 cm-250 cm underwater and reacts to catch the fish. In this study, studies were carried out on finding the appropriate trap modeling for catching the target pufferfish, developing a suitable model for catching the target fish species, and verifying the target species recognition by using different artificial intelligence methods.