Deep Learning-Based Image Recognition for Soft-Shell Shrimp Classification

TL;DR

This study tackles automatic classification of soft-shell versus hard-shell white shrimp immediately after harvest to address shell integrity and visual quality concerns. It proposes a CNN-based image recognition pipeline trained on high-resolution scans of live shrimp captured with an ApeosPort scanner to substitute manual sorting. The dataset comprises 0131Dataset-7 with 398 ordinary and 84 soft-shell shrimp, obtained via expert labeling and 80/20 train/validation split, with data augmentation to balance classes. The model achieved approximately 87% validation accuracy after 30 epochs, demonstrating feasibility for real-world shrimp sorting and potential improvements through dataset expansion and hybrid classifiers.

Abstract

With the integration of information technology into aquaculture, production has become more stable and continues to grow annually. As consumer demand for high-quality aquatic products rises, freshness and appearance integrity are key concerns. In shrimp-based processed foods, freshness declines rapidly post-harvest, and soft-shell shrimp often suffer from head-body separation after cooking or freezing, affecting product appearance and consumer perception. To address these issues, this study leverages deep learning-based image recognition for automated classification of white shrimp immediately after harvest. A convolutional neural network (CNN) model replaces manual sorting, enhancing classification accuracy, efficiency, and consistency. By reducing processing time, this technology helps maintain freshness and ensures that shrimp transportation businesses meet customer demands more effectively.

Figures (4)

• Figure 1: Comparison of Hard-Shell Shrimp (Top) and Soft-Shell Shrimp (Bottom)
• Figure 2: The Research Architecture
• Figure 3: The Process of Risk Assessment and Financial Analysis
• Figure 4: 0131Dataset-7 Training Curve Chart (Left) and Validation Curve Chart (Right)