A Deep Learning Approach to Recognize Bangladeshi Shrimp Species

Abstract

The nutritional value of shrimp, the most consumed shellfish in Bangladesh, is high because of its abundance of various nutrients. Protein, minerals, vitamin D, and iodine are all examples of such things. The name of this seafood means "white gold" in its native Bangladeshi language. More than 70% of the world's agricultural food supply comes from this. Roughly 56 distinct species of shrimp can be found in American waters. Unfortunately, most people only have a cursory understanding of the wide variety of life on our planet. Experienced fishermen sometimes make the mistake of confusing one species with another because of their superficial resemblance. We developed an AI system to aid in shrimp identification to solve the difficulty presented by this paper. We anticipate that this study will also help the export sector in distinguishing between the numerous shrimp species under observation. To accomplish our goals, we developed an in-house convolutional neural network (CNN) algorithm for extracting features and processing images. Here, we construct three unique CNN architectures, each with its own unique hyperparameters and convolutional layers. Model 3 was chosen as the final model for computer vision integration, despite the fact that both Model 1 and Model 3 attained an accuracy of 99.01%. Models 1 and 3 produce the same answer; if model 3 is more accurate, why would we pick it as the final model? Furthermore, we will offer a rational justification for this study's findings.