NOTION - Net hOle deTectION

NOTION project deals with one of the main challenges the aquaculture industry is facing today: escapes from fish cages. To avoid or limit the impacts of such events a ROV mounted subsystem is proposed and designed. This subsystem can detect the damaged nets rapidly before escapes can occur. The tool's core system is advanced image processing algorithms for the automated detection of damaged net (holes). The use of smaller ROVs is growing in the aquaculture industry and this project proposes state-of-the art technology so as to eliminate human errors.

Οbjective of NOTION project is to scan fishing nets, detect all damages (holes) of the net and discriminated them from the normal net’s holes. As a “damaged area” is defined any opening in the nets structure that is bigger than the 90% of all other holes present in the image.

To achieve both these objectives, an extensive image processing procedure had to be developed, that could:

1. Analyse in real time the incoming video stream
2. Detect every hole of the net (normal or damage)
3. Calculate dimensions of every hole and if it’s bigger than the 90% of the rest present, classify it as a damaged area.
4. Check if the damaged area is present for a number of seconds (to exclude possibility of false detection due to noise etc)
5. Track the detected damaged area
6. Calculate from the image the position of the area