Publication details
Artificial neural networks for insect identification.
| Authors | |
|---|---|
| Year of publication | 2006 |
| Type | Article in Proceedings |
| Conference | Proceedings of the 6th International Congress of Dipterology |
| MU Faculty or unit | |
| Citation | |
| Field | Zoology |
| Keywords | ANN Insects Identification |
| Description | ANN have already been very rarely applied for identifying of insects. We developed, tested and applied the ANN methodology at three different insect orders: I. Diptera-Tachinidae: There were 17 characters studied, particularly in males and females (Tachina), or in right or left wings (Ectophasia). Additionally, the sex of the studied specimens was included. It was shown that classification using ANN was possible assuming sufficiently high number of specimens of each species in the training set. II. Thysanoptera-Thripidae: The three species were characterized by 20 quantitative and three binary characters, and by sex. Thus, reliable characters were found and the model provided a fast identification tool. III. Hemiptera-Psylloidea-Phacopteronidae: Adults of up to ten species, 17 characters on forewings, antenna, head and hind legs, and sex were used. The ANN model performed well on the data set and unambiguously classified unknown samples. |
| Related projects: |