Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Lightweight Fish Classification Model for Sustainable Marine Management: Indonesian Case

Febrian Kurniawan, Gandeva Bayu Satrya, Firuz Kamalov

TL;DR

The paper tackles sustainable fisheries management by enabling on-board, real-time fish species and consumability classification using a lightweight CNN, M-MobileNets, designed for resource-constrained vessels. It introduces a large Indonesian fish image dataset (37,462 images across 667 species) and data augmentation, then demonstrates a two-stage approach that first identifies species and then determines consumability via FishBase integration. By reducing the top-layer parameters to about 531k and adopting Swish activation with batch normalization, the model achieves up to 97% accuracy and favorable GPU/memory metrics on a low-end GPU (GTX 860M), enabling practical edge deployment. The work supports synchronized multi-vessel deployments for monitoring species distribution and movement, with clear implications for remote, low-bandwidth marine AI applications and sustainable fishing practices.

Abstract

The enormous demand for seafood products has led to exploitation of marine resources and near-extinction of some species. In particular, overfishing is one the main issues in sustainable marine development. In alignment with the protection of marine resources and sustainable fishing, this study proposes to advance fish classification techniques that support identifying protected fish species using state-of-the-art machine learning. We use a custom modification of the MobileNet model to design a lightweight classifier called M-MobileNet that is capable of running on limited hardware. As part of the study, we compiled a labeled dataset of 37,462 images of fish found in the waters of the Indonesian archipelago. The proposed model is trained on the dataset to classify images of the captured fish into their species and give recommendations on whether they are consumable or not. Our modified MobileNet model uses only 50\% of the top layer parameters with about 42% GTX 860M utility and achieves up to 97% accuracy in fish classification and determining its consumability. Given the limited computing capacity available on many fishing vessels, the proposed model provides a practical solution to on-site fish classification. In addition, synchronized implementation of the proposed model on multiple vessels can supply valuable information about the movement and location of different species of fish.

Lightweight Fish Classification Model for Sustainable Marine Management: Indonesian Case

TL;DR

The paper tackles sustainable fisheries management by enabling on-board, real-time fish species and consumability classification using a lightweight CNN, M-MobileNets, designed for resource-constrained vessels. It introduces a large Indonesian fish image dataset (37,462 images across 667 species) and data augmentation, then demonstrates a two-stage approach that first identifies species and then determines consumability via FishBase integration. By reducing the top-layer parameters to about 531k and adopting Swish activation with batch normalization, the model achieves up to 97% accuracy and favorable GPU/memory metrics on a low-end GPU (GTX 860M), enabling practical edge deployment. The work supports synchronized multi-vessel deployments for monitoring species distribution and movement, with clear implications for remote, low-bandwidth marine AI applications and sustainable fishing practices.

Abstract

The enormous demand for seafood products has led to exploitation of marine resources and near-extinction of some species. In particular, overfishing is one the main issues in sustainable marine development. In alignment with the protection of marine resources and sustainable fishing, this study proposes to advance fish classification techniques that support identifying protected fish species using state-of-the-art machine learning. We use a custom modification of the MobileNet model to design a lightweight classifier called M-MobileNet that is capable of running on limited hardware. As part of the study, we compiled a labeled dataset of 37,462 images of fish found in the waters of the Indonesian archipelago. The proposed model is trained on the dataset to classify images of the captured fish into their species and give recommendations on whether they are consumable or not. Our modified MobileNet model uses only 50\% of the top layer parameters with about 42% GTX 860M utility and achieves up to 97% accuracy in fish classification and determining its consumability. Given the limited computing capacity available on many fishing vessels, the proposed model provides a practical solution to on-site fish classification. In addition, synchronized implementation of the proposed model on multiple vessels can supply valuable information about the movement and location of different species of fish.
Paper Structure (18 sections, 10 equations, 7 figures, 3 tables)

This paper contains 18 sections, 10 equations, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Remote deployment
  3. Communication Technology
  4. Hardware Resources
  5. Consumability Determination
  6. Fish Classification Model
  7. Data
  8. Data Augmentation
  9. Mechanism
  10. MobileNets
  11. Modified MobileNets
  12. Categorization
  13. Transfer Learning
  14. Evaluation and Analysis
  15. Performance Evaluation
  16. ...and 3 more sections

Figures (7)

  • Figure 1: Comparison of the number of parameters in the top layer for MobileNets and M-MobileNets.
  • Figure 2: The architecture design of M-MobileNets.
  • Figure 3: Example of final classification result: Consumable
  • Figure 4: Example of final classification result: Unconsumable
  • Figure 5: Activation functions considered in the study.
  • ...and 2 more figures