MATANet: A Multi-context Attention and Taxonomy-Aware Network for Fine-Grained Underwater Recognition of Marine Species

TL;DR

MATANet tackles fine-grained underwater species recognition by integrating surrounding habitat context and taxonomic hierarchy. It introduces MCEAM to model ROI-context interactions and HSLM to embed taxonomy through level-wise auxiliary classifiers, optimized with $L_{total}=L_{cls}+L_{hier}$. The approach achieves state-of-the-art results on FathomNet2025, FAIR1M, and LifeCLEF2015-Fish datasets, supported by ablations and visualizations showing hierarchical consistency and meaningful ROI-context interactions. This work enables more reliable automated biodiversity monitoring by combining contextual reasoning with hierarchical, taxonomic guidance.

Abstract

Fine-grained classification of marine animals supports ecology, biodiversity and habitat conservation, and evidence-based policy-making. However, existing methods often overlook contextual interactions from the surrounding environment and insufficiently incorporate the hierarchical structure of marine biological taxonomy. To address these challenges, we propose MATANet (Multi-context Attention and Taxonomy-Aware Network), a novel model designed for fine-grained marine species classification. MATANet mimics expert strategies by using taxonomy and environmental context to interpret ambiguous features of underwater animals. It consists of two key components: a Multi-Context Environmental Attention Module (MCEAM), which learns relationships between regions of interest (ROIs) and their surrounding environments, and a Hierarchical Separation-Induced Learning Module (HSLM), which encodes taxonomic hierarchy into the feature space. MATANet combines instance and environmental features with taxonomic structure to enhance fine-grained classification. Experiments on the FathomNet2025, FAIR1M, and LifeCLEF2015-Fish datasets demonstrate state-of-the-art performance. The source code is available at: https://github.com/dhlee-work/fathomnet-cvpr2025-ssl

Figures (4)

• Figure 1: Different visual contexts for marine species recognition. The ROI image shows a sea anemone (family: Hormathiidae). Its taxonomic identity becomes more evident when the surrounding organisms and habitat features visible in the context and full-context images are taken into account.
• Figure 2: Overview of MATANet. The model processes ROI and multi-scale contextual images using ViT embeddings. The Multi-Context Environmental Attention Module (MCEAM) applies cross-attention between the ROI and contextual regions to integrate their features. The Hierarchical Separation-Induced Learning Module (HSLM) promotes taxonomic consistency through level-specific auxiliary classifiers.
• Figure 3: t-SNE visualization of embeddings extracted from MCEAM under different HSLM settings. Subfigures (a), (b), and (c) correspond to ablation study models M3 (MCEAM without HSLM), M6 (MCEAM with HSLM–Random), and M4 (MCEAM with HSLM), respectively. The model with HSLM (c) exhibits more taxonomically consistent embedding distributions, demonstrating the effectiveness of HSLM in preserving hierarchical structure.
• Figure 4: Cross-attention between ROI and multi-scale context regions in MCEAM. Each row shows attention from the ROI to ×3, ×5, and full-image contexts. MATANet captures distinct patterns such as separated attention, complementary attention, and selective attention, focusing on habitat-relevant features while ignoring irrelevant ones.