MobileNetV2 Hyperparameter Optimization for Freshness Classification of Tuna Fish Based on Fish Eye Imagery Using Optuna Tree-structured Parzen Estimator (TPE)

Wa Ode Asriyani; Muhammad Ihsan Sarita; Sutardi Sutardi

doi:10.31004/riggs.v5i1.11506

Authors

Wa Ode Asriyani Universitas Halu Oleo
Muhammad Ihsan Sarita Universitas Halu Oleo
Sutardi Sutardi Universitas Halu Oleo

DOI:

https://doi.org/10.31004/riggs.v5i1.11506

Keywords:

Tuna, Fish Freshness, MobileNetV2, Optuna, Transfer Learning, Tree-structured Parzen Estimator (TPE)

Abstract

Manually assessing the freshness of tuna (Euthynnus affinis) within traditional auction markets is frequently subjective, inconsistent, and prone to human error. To address these limitations, this study implemented a robust Deep Learning model utilizing the MobileNetV2 architecture to classify fish freshness levels based on ocular imagery into three distinct categories: Fresh, Less Fresh, and Not Fresh. Although the MobileNetV2 architecture is inherently reliable for image classification, conventional manual hyperparameter tuning often leads to suboptimal model configurations and severe overfitting challenges. As a strategic solution, this study implemented the advanced Tree-structured Parzen Estimator (TPE) algorithm through the Optuna framework to automatically optimize critical hyperparameters, including learning rate, dropout rate, optimizer type, and the fine-tune depth of the convolutional layers. Empirical testing results rigorously proved that the strategic integration of Optuna TPE with the MobileNetV2 model successfully provided a significant performance boost. The optimized model achieved an accuracy of 93.51%, representing a substantial increase of 32.04% compared to the 61.47% accuracy observed in the manual baseline model. Correspondingly impressive improvements were recorded in key performance metrics, specifically achieving a Precision of 93.64%, a Recall of 93.37%, and an F1-Score of 93.49%. In conclusion, this research demonstrates that automated hyperparameter optimization using the Optuna TPE framework significantly enhances the reliability, efficiency, and generalization capability of Deep Learning models, establishing a highly accurate and practical solution for the automated detection of tuna freshness in real-world fishery environments.

Downloads

Download data is not yet available.

References

[1] H. Fitriyah , D. Syauqy , and FA Susilo, “ Detection Freshness of Mackerel tuna ( Euthynnus) Affinis ) Automatically Based on Eye Images Using Binary Similarity,” Journal of Information Technology and Computer Science , vol. 7, no. 5, pp. 879–886, Oct. 2020, doi : 10.25126/jtiik.2020753839.

[2] MFN Al Fatich , AI Setyastuti , D. Kresnasari , and S. Sarmin , " Identification of Freshness Level of Mackerel tuna ( Euthynnus sp.) at Bumiayu Market , Bumiayu Regency Brebes ,” J. Mar. Res. , vol. 12, no. 3, pp. 511–518, Jul. 2023, doi : 10.14710/ jmr.v 12i3.40444.

[3] M. Muchtar , YP Pasrun , R. Rasyid , N. Miftachurohmah , and M. Mardiawati , " Application of the Naïve Bayes Method in Classification Freshness of Fish Based on Color in Eye Area Images,” Journal of Informatics and Electrical Engineering Applied , vol. 12, no. 1, Jan. 2024, doi : 10.23960/ jitet.v 12i1.3879.

[4] VT Deco, A. Nilogiri , and Q. A'yun , “ Classification The Freshness of Lemuru Fish Based on Eye Images Using Convolutional Neural Network Method . , ” Journal of Information Systems and Electronics Applications , vol. 3, no. 1, pp. 14–25, 2021.

[5] IR Alamsyah , SN Budiman, and U. Mawaddah , " Implementation of Deep Learning Using Convolutional Neural Network (CNN) for Classification of Goldfish Types (Carassius Auratus) Based on Characteristics Morphology .,” Infomedia Journal : Informatics Engineering , Multimedia, and Networks , vol. 10, no. 2, pp. 119–124, 2025.

[6] MAK Raiaan et al. , “A systematic review of hyperparameter optimization techniques in Convolutional Neural Networks,” Decision Analytics Journal , vol. 11, p. 100470, Jun. 2024, doi : 10.1016/j.dajour.2024.100470.

[7 ] X. Liu , Y. Hu , 15, no. 1, p. 18, Dec. 2024, doi : 10.3390/agronomy15010018.

[8] WON Emelia, I. Irawati , and N. Kurniati , “ Classification Freshness of Lencam Fish ( Lethrinus) Atkinsoni Seale) Using the K-Nearest Neighbor Method Based on Digital Imagery,” Bulletin of Islamic Information Systems and Technology , vol. 4, no. 3, pp. 274–285, Aug. 2023, doi : 10.33096/ busiti.v 4i4.1778.

[9] G. Rong et al. , “Comparison of Tree-Structured Parzen Estimator Optimization in Three Typical Neural Network Models for Landslide Susceptibility Assessment,” Remote Sens. (Basel). , vol. 13, no. 22, p. 4694, Nov. 2021, doi : 10.3390/rs13224694.

[10] RR Santosa , A. Sasikirana , ZL Ramadhan, YPB Satria, and A. Nugroho, " Performance Comparison of MobileNetV2 and ResNet50V2 in Maturity Level Classification Tomatoes ," Indonesian Journal of Science, Technology and Humanities , vol. 3, no. 1, pp. 49–56, Jun. 2025, doi : 10.60076/ ijstech.v 3i1.1335.

[11] YY Pane and JJ Sihombing , “ Bird Species Classification Using the Transfer Learning Method,” Journal of Technology Integrated , vol. 9, no. 2, pp. 89–94, Dec. 2023, doi : 10.54914/ jtt.v 9i2.744.

[12] S. Sieradzki and J. Mańdziuk , “Modified Adaptive Tree-Structured Parzen Estimator for Hyperparameter Optimization,” Feb. 2025.

[13] G. Hariharan, R. Mallya, N. Kumar, D. Doreswamy , GM Chennegowda , and SK Bhat, “ Optuna -Optimized Ensemble and Neural Network Models for Static Characteristics Prediction of Active Bearings with Geometric Adjustments,” Modeling , vol. 6, no. 3, p. 98, Sept. 2025, doi : 10.3390/modelling6030098.

[14] K. Abe, Y. Wang, and S. Watanabe, “Tree-Structured Parzen Estimator Can Solve Black-Box Combinatorial Optimization More Efficiently,” Jul. 2025.

[15] R. Kochnev , A.T. Goodarzi , Z.A. Bentyn , D. Ignatov, and R. Timofte , “ Optuna vs Code Llama: Are LLMs a New Paradigm for Hyperparameter Tuning?, ” Sep. 2025.

[16] FM Fathoni , CA Putra, and AL Nurlaili , “ Classification of Grape Leaf Diseases Using the K-Nearest Neighbor Method Based on the Gray Level Co-Occurrence Matrix,” Biner: Scientific Journal Informatics and Computers , vol. 3, no. 1, pp. 8–15, Jan. 2024, doi : 10.32699/biner.v3i1.6332.

[17] RA Kumala , CA Sari, and EH Rachmawanto , "A Comparison of MobileNetV2 and VGG16 Architectures with Transfer Learning for Multi-Class Image-Based Waste Classification," Journal of Applied Informatics and Computing , vol. 9, no. 4, pp. 1610–1624, Aug. 2025, doi : 10.30871/ jaic.v 9i4.9958.