Automatization of the insect
recognition process

About system

 

T he main goal of IPM is to protect the cultural heritage collections against insects. The proper and successful monitoring results in the correct identification of the caught insects.

Insect classification in terms of species and quantity is a source of knowledge about the ecosystem. Based on this knowledge, one can determine long-term statistical trends, which allow to assess the nature and the scale of the hazard.

When insect classification is automated, the IPM coordinator can focus on the risk analysis instead of performing labour-intensive tasks.

Recognition process

 

N euronic® is a specialized computer application created for museum coordinatiors and other people who supervise IPM-based systems. It is intended for automatic classification of insects caught in the monitoring traps.

Neuronic creates a new standard in IPM. It enables to digitize and thus to archive source data from IPM procedures. It allows to save high-resolution photos of all insects caught in monitoring traps.

Neuronic® performs insect recognition by analyzing trap images, taken with the smartphone. The app works with selected Samsung models.

The whole data available on the smartphone digital matrix, or the so-called raw files, are analyzed. The analysis is performed using an appropriately constructed neural network.

The QR code database allows to properly mark the set traps in a manner that the app can recognize.

Work with Neuronic

 

T he IPM coordinator takes photos of monitoring traps using:
-smartphone
-Hemisphere – “Beetle”

The IPM coordinator may review data using a computer and decide on the further analysis. Neuronic can export data to Excel
files.

The Neuronic app uses neural networks. Each stage is performed using a different neural network. Neural networks are the structures which use various mathematical calculations in their operation.

Classification efficiency

 

N euronic uses several dozen million operation types to find dependencies, which allow to categorize the specific pictures into particular groups. During the learning procedure, multiple repetitive tests are performed.

They verify the learned capability to work independently. The tests consist in verifying the network with hitherto unknown pictures. Within the testing procedure framework, trap images, described in details by experienced entomologists, undergo analysis conducted by the network. The test is highly strenuous for the network.

ANY QUESTIONS? BEST CHECK AT THE SOURCE.

SEE THE PRESENTATION

E-mail

maria@panko.pl, bartlomiej@panko.pl

ADDRESS

SET Bartłomiej Pankowski ul. Polna 9G 05-500 Mysiadło

Phone number

+48 603 929 989
+48 603 771 074