Bangladeshi Shrimp Species Recognition By Deep Convolutional Neural Network

Abstract

Shrimp, the most popular shellfish in Bangladesh, is rich in a variety of nutrients, giving it a great nutritional value. Some examples of such substances include iodine, protein, minerals, and vitamin D. This shellfish is named from the Bangladeshi word for "white gold" in its own language. It is the source of almost 70% of the agricultural food that the world consumes. The seas of Bangladesh are home to over 56 different species of prawn. Even seasoned fishers can't help but mix up different species when they see a superficial similarity. The authors of this research set out to address the problem of prawn identification, thus I built an AI system to help. I achieved our objectives by creating our own convolutional neural network (CNN) method for feature extraction and picture processing. In this work, I build three distinct convolutional neural network (CNN) designs, with distinct hyperparameters and convolutional layers for each, and one transfer learning approach vtgg19 is used. I got the highest accuracy from model 2 custom CNN architecture. The accuracy rate is 99.14%.