Answers are grounded in this paper and cite their sources.
The complexity and opacity of biological data in fisheries makes the deep application of artificial intelligence in marine fishery management challenging. Moreover, establishing a reliable model for fish identification is vital for the management of fishery biological data, and interpretable methods enable us to deeply understand the decision-making mechanism of such models. This study selected Thunnus albacares as the research target and constructed an image classification framework based on the VGG16 model. Advanced techniques such as feature map visualization were used to analyze the decision-making mechanism of this model. The experimental data included standard samples, challenging samples, and samples obtained under various noisy conditions. This study revealed the hierarchical fusion process of the model, indicating that the key decision-making areas of the model highly coincide with the identification marks used in traditional biology. The results of this study will provide a solid foundation for improving related models and strongly promote the in-depth development of artificial intelligence in the fields of fishery biology and fishery management.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: