Fish Inspection System Using a Parallel Neural Network Chip and the Image Knowledge Builder Application

• Main Authors: Menendez, Anne, Paillet, Guy
• Format: Article in Journal/Newspaper
• Language: English
• Published: Association for the Advancement of Artificial Intelligence 2008
• Subjects: Norway; Iceland
• Online Access: https://ojs.aaai.org/index.php/aimagazine/article/view/2084; https://doi.org/10.1609/aimag.v29i1.2084

A generic image learning system, CogniSight, is being used for the inspection of fishes before filleting offshore. More than 30 systems have been deployed on seven fishing vessels in Norway and Iceland over the past three years. Each CogniSight system uses four neural network chips (a total of 312 neurons) based on a natively parallel, hard-wired architecture that performs real-time learning and nonlinear classification (RBF). These systems are trained by the ship crew using Image Knowledge Builder, a ”show and tell” interface that facilitates easy training and validation. Fishers can reinforce the learning anytime when needed. The use of CogniSight has significantly reduced the number of crew members needed on the boats (by up to six persons), and the time at sea has been shortened by 15 percent. The fast and high return of investment (ROI) to the fishing fleet has significantly increased the market share of Pisces Industries, the company integrating CogniSight systems to its filleting machines.