Monitoring of the flight of olive kernel borer by developing and using model based on artificial neural networks

Maslinin moljac (Prays oleae Bernard, 1788) jedan je od najznačajnijih štetnika masline. U maslinicima radi štete velikih gospodarskih razmjera, najčešće u obliku opadanja plodova, slabijem urodu i sporijem rastu stabla. Kako bi se njegova prisutnost na terenu ranije uočila, predložen je razvoj modela za automatsku detekciju temeljen na prepoznavanju sadržaja slika pomoću umjetnih neuronskih mreža. Postavljanjem lovki s feromonima, prikupljanjem i fotografiranjem uzoraka, na lovkama ulovljenih, odraslih oblika P. oleae na ljepljivim pločama, anotiranjem fotografija i učenjem neuronske mreže, razvijen je model za automatsko praćenje ovog štetnika. Osim vrste P. oleae, model je naučen prepoznavati ostale kukce kao i ostale elemente na slikama. Preciznost prepoznavanja odraslih oblika moljca iznosila je 59 % u odnosu na vizualni pregled, a veća je od preciznosti prepoznavanja ostalih štetnika i ostalih elemenata na fotografijama ljepljivih ploča.Olive kernel borer (Prays oleae Bern ard, 1788) is one of the most significant pests of olive tree. In olive groves, it can cause damage of large economic proportions , commonly in form of fruit drop, weaker yield, or stagnant tree growth . To detect its presence in the field as soon as possible, an automatic model based on the image recognition of image content using artificial neural networks is used. By setting up phe romone based traps, collecting and photographing samples of adult P. oleae, annotating photographs and teaching a neural network to use them, an automatic model is developed for tracking olive kernel borer individuals. Besides the species P. oleae , the mod el was taught to recognize other pests as well as other elements of proposed images. The accuracy of recognition of the adult moths was 59 compared to visual inspection of proposed photographs and was higher that the accuracy of recognition of other pest s and other elements on photographs of adhesive panels