A Deep Learning-based Approach for Shrimp Quality Estimation using DenseNet 121

• Main Authors: J, Bipin Nair B, P, Akhil, N, Shashank M
• Format: Conference Proceeding
• Language: English
• Subjects: Classification; Computer architecture; DenseNet121; Estimation; Industries; Neural Network; Process control; Quality assessment; Quality control; Quality Estimation; Shiba Shrimp; Shrimp; Training
• ISSN: 2767-7788
• DOI: 10.1109/ICICT60155.2024.10544780

Shrimp creates an immense significance in the global scale due to their nutritional and cultural importance. They provide plentiful supply of amino acids, vital nutrients, omega-3 fatty acids and many more. The extensive challenge of obtaining a dataset and efficiently classifying it based on the quality remains an ongoing area of research. The manually collected dataset was expanded to increase the generalization ability of the network. The model achieved an impressive remarkable results attaining an accuracy of 95.03% during training and 98.75% on the testing dataset, highlighting its efficiency and effectiveness. The classification was performed across three categories of shrimp dataset: Fresh, Frozen and Stale shrimp images that were manually collected consisting of 5800 images in total. This work contributes to the enhancement in quality control of shrimps, benefiting industries that depends efficient classification processes.