Skip to content
Open access

A comparative evaluation of deep learning architectures, loss functions, and ensemble strategies for underwater fish species classification

Aug 2026 · Scientific Reports · 0 citations

TL;DR

The experimental results indicate that class-weighted cross-entropy loss consistently provided stronger performance for individual models, whereas focal loss produced a marginal improvement when used with the soft-voting ensemble.

Abstract

The accurate classification of underwater fish species is important for biodiver-sity monitoring, sustainable fisheries management, and aquaculture. However, underwater fish classification remains challenging due to class imbalance, varying imaging conditions, and high visual similarity among species. This study presents a comprehensive evaluation of deep learning architectures, loss functions, and ensemble strategies for underwater fish species classification. Using the Mark Daniel Lampa Kaggle Fish Dataset comprising 13,304 images across 31 species, duplicate images were removed using a combination of MD5 hashing and perceptual hashing, resulting in a cleaned dataset for model training and evaluation. EfficientNet-B0, EfficientNetV2-S, Swin-Tiny Transformer, and ConvNeXt-Tiny were trained using both class-weighted cross-entropy loss and focal loss. The experimental results indicate that class-weighted cross-entropy loss consistently provided stronger performance for individual models, whereas focal loss produced a marginal improvement when used with the soft-voting ensemble. Among the individual models, EfficientNet-B0 achieved the highest accuracy of 94.81%. Ensemble learning further improved performance, with the four-model soft-voting ensemble achieving a test accuracy of 97.21% and a macro-F1 score of 0.9678 on the cleaned Kaggle dataset. An external dataset of 310 internet-sourced fish images was additionally used to assess model generalization under domain-shift conditions, where the ensemble achieved an accuracy of 76.39% and a macro-F1 score of 0.7335. While performance decreased substantially under domain shift, the results highlight the challenges of generalizing fish-species classification models beyond the training distribution. Fish species classification represents an important component of future end-to-end underwater monitoring systems incorporating fish detection and localization modules. The study provides a benchmark for future research on underwater fish-species classification, ensemble learning, and real-world deployment. The source code is openly available on GitHub and permanently archived on Zenodo (DOI: https://doi.org/10.5281/zenodo.19315919 ).

Read PDF

Similar papers

Open access Aug 2026

DEEPFIN: A deep learning tool for fish image classification from unlabeled data

DEPFIN is an annotation-free image analysis pipeline that combines foreground segmentation, frozen pretrained convolutional encoders, non-linear dimensionality reduction, and density-based clustering for extracting biological structure from the growing volumes of unlabeled fish imagery.

Alexandru Mihai, Billy Moore, Marleen Klann et al. · 0 citations
Open access Oct 2026

Spatial attention–driven hybrid Vision Transformers for automated classification of crown-of-thorns starfish in coral reef ecosystems

Coral reefs play the main role in marine ecology, as home and shelter to many aquatic species, and quite importantly, supporting local economies. The danger to coral reefs from coral bleaching is joined by Acanthaster planci invasions that can extensively harm these sensitive colonies. The normal practice has been...

Y. P, M. P, V. K. · 0 citations
Open access Sep 2026

FishMonitorAI: An Efficient and Lightweight Deep Learning Model for Underwater Fish Detection in Resource-Constrained Environments

Accurate and efficient fish detection in underwater environments is fundamental to the advancement of automated aquaculture monitoring and selective fishing systems. However, the deployment of existing object detection models in such environments remains challenging due to wavelength-dependent light absorption, suspend...

Van Le, M. Astapova, M. Uzdiaev et al. · 0 citations
Open access Sep 2026

JPNet: A multi-layered fusion deep learning architecture for the detection and classification of pests in jute crops

Detecting and classifying insect pests is a critical challenge in agricultural pest management, as infestations can reduce crop yield and quality. This study introduces JPNet, a convolutional neural network (CNN) architecture that uses multi-layer feature fusion to detect and classify insect pests affecting jute crops....

Mejbah Ahammad, Md. Ashraful Babu, M. Hossain et al. · 0 citations
Open access Aug 2026

A Deep Residual Learning Framework using ResNet-152 for Accurate Multi-class Classification of Shrimp Diseases

Background: Shrimp aquaculture is a major contributor to global seafood production. Disease outbreaks remain a serious challenge, causing severe economic losses and threatening food security. Timely and accurate disease diagnosis is essential for effective farm management. However, conventional diagnostic methods are l...

Kil-hwan Shin · 0 citations
Open access Sep 2026

YOLO-CBNet: A robust attention-enhanced detection framework for underwater fish recognition in aquaculture environments

The accurate identification of fish species remains a pivotal challenge in aquaculture, as it directly influences population management, health monitoring, and overall production efficiency. Yet, underwater environments often present difficult conditions such as reduced visibility, suspended particles, and uneven light...

Mahdi Hamzaoui, Leila Bousbia, Mohamed Ould-Elhassen Aoueileyine et al. · 1 citation

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.