Deep EfficientNetB3 Framework for Automated Classification of Marine Animals from Underwater Imagery
Keywords:
Underwater image classification, Marine animal recognition, Fish species classification, EfficientNetB3, Transfer learning, Deep learning, Computer vision, Marine biodiversity monitoring, Automated species identificationAbstract
The terrible increase of under water imaging technologies has opened new fields for marine biodiversity nowadays automated monitoring, however, accurately classification of marine species is actually quite challenging taking into account complex visual conditions like illumination variation, occlusion and background clutter. This study provides a deep EfficientNetB3-based framework for the automated classification of marine animals from the underwater images. Leveraging Transfer Learning with ImageNet Pre-Trained Weights The proposed model is fine-tuned to successfully extract discriminative features specific to underwater fish species using ImageNet pre-trained weights. A public accessible dataset taken from the Kaggle platform is used to test the framework under realistic conditions. Comprehensive experiments show that such an approach is able to reach a 98% accuracy of the classification, outperforming several state-of-the-art deep learning architectures. Detailed performance analysis using precision, recall, F1 score, confusion matrix and ROC-AUC metrics provides additional validations to robustness and generalization capacity model with AUC value 0.99. The results speak to the usefulness and accuracy-computing efficiency of EfficientNetB3. The proposed framework presents a stable and scalable solution for the automated recognition of marine species, which has applications in marine ecology, biodiversity assessment, and advanced underwater monitoring systems.
References
[1] Ergün, E., 2025. SwinFishNet: A Swin Transformer-based approach for automatic fish species classification using transfer learning. PLoS One, 20(5), p.e0322711.
[2] Zhou, P., He, X.Q., Tu, Z.Y., Sun, D., Wang, C.S., Shen, H.B. and Pan, X., 2025. Bridging the gap between the public’s knowledge and detection of marine non-indigenous species through developing automated image classification applications for marine species. Frontiers in Marine Science, 12, p.1508851.
[3] Jamala, M.N., Al Deeb, M. and Abu-Naser, S.S., 2025. Identifying Fish Species Using Deep Learning Models on Image Datasets.
[4] Ziaei, M.M., Urban, J. and Cisar, P., 2025. Image‐Based Individual Fish Identification as a Substitute for Invasive Methods: State‐of‐the‐Art and Research Gaps. Reviews in Aquaculture, 17(4), p.e70078.
[5] Jerez-Tallón, M., Beviá-Ballesteros, I., Garcia-D’Urso, N., Toledo-Guedes, K., Azorín-López, J. and Fuster-Guilló, A., 2025, June. Comparative Study of Deep Learning Approaches for Fish Origin Classification. In International Work-Conference on Artificial Neural Networks (pp. 66–78). Cham: Springer Nature Switzerland.
[6] Kumar, P. and Shukla, P.K., 2025. Hybrid transformer-CNN model for precise fish segmentation, localization, and species classification in aquaculture. Journal of Integrated Science and Technology, 13(6), pp.1134–1134.
[7] Córdova, M., Torres, R.D.S., Helmond, A.V. and Kootstra, G., 2025. Novelty Recognition: Fish Species Classification via Open-Set Recognition. Sensors, 25(5), p.1570.
[8] Martins, J.R.D.S., 2025. Fish Species Classification in Controlled Underwater Environments Using Contrastive Language-Image Models. Doctoral dissertation.
[9] Kant, V. and Yamsani, N., 2025, April. Animal Species Classification Using Fine-Tuned ResNet152V2: A Deep Learning Approach for Biodiversity Conservation. In 2025 3rd International Conference on Advancement in Computation & Computer Technologies (InCACCT) (pp. 35–39). IEEE.
[10] Tejaswini, H., Pai, M.M. and Pai, R.M., 2024. Automatic estuarine fish species classification system based on deep learning techniques. IEEE Access.
[11] Yang, C., Zhou, P., Wang, C.S., Fu, G.Y., Xu, X.W., Niu, Z., Zhu, L., Yuan, Y., Shen, H.B. and Pan, X., 2024. FishAI: Automated hierarchical marine fish image classification with vision transformer. Engineering Reports, 6(12), p.e12992.
[12] Jareño, J., Bárcena-González, G., Castro-Gutiérrez, J., Cabrera-Castro, R. and Galindo, P.L., 2024. Automatic labeling of fish species using deep learning across different classification strategies. Frontiers in Computer Science, 6, p.1326452.
[13] Mohammadisabet, A., Hasan, R., Dattana, V., Mahmood, S. and Hussain, S., 2025. CNN-Based Optimization for Fish Species Classification: Tackling Environmental Variability, Class Imbalance, and Real-Time Constraints. Information, 16(2), p.154.
[14] Malik, H., Naeem, A., Hassan, S., Ali, F., Naqvi, R.A. and Yon, D.K., 2023. Multi-classification deep neural networks for identification of fish species using camera captured images. PLoS One, 18(4), p.e0284992.
[15] Gong, B., Dai, K., Shao, J., Jing, L. and Chen, Y., 2023. Fish-TViT: A novel fish species classification method in multi water areas based on transfer learning and vision transformer. Heliyon, 9(6).
[16] Alsmadi, M.K. and Almarashdeh, I., 2022. A survey on fish classification techniques. Journal of King Saud University – Computer and Information Sciences, 34(5), pp.1625–1638.