Sistema de reconocimiento y clasificación de camarón utilizando visión artificial

Abstract

The present project” Shrimp recognition and classification system using artificial vision” proposes an innovative solution to traditional shrimp classification methods, replacing manual classification with automatic identification of their main characteristics through the implementation of new technologies such as computer vision and neural networks. For the implementation of the prototype, a hopper is available where the shrimps are deposited and transported to a conveyor belt with better distribution. Next, the shrimps pass under a box where a camera using an artificial vision system programmed on a Raspberry Pi is located. This system allows for digital classification of the shrimps according to their distinct characteristics and features. All of this is carried out in a controlled environment with good lighting. Finally, a neural network was used, and various tests were carried out by adjusting the weights, number of layers, and epochs to obtain a model that would accurately recognize the characteristics and features of the shrimp. After carrying out several field tests, a model has been obtained that has an accuracy of 93% in shrimp recognition. This means that the model is capable of correctly identifying the shrimp in most cases and has a high capacity for precision in its predictions.