<![CDATA[Automatic identification of fish species plays an important role in various fields such as conservation biology, fisheries management, and biological research. The Convolutional Neural Network (CNN) method has become an effective solution for automating this process using digital images. However, achieving high accuracy requires careful consideration of factors such as the quantity and quality of data, image preprocessing methods, feature extraction techniques, classification algorithms, and optimization strategies. This study addresses these challenges by proposing a CNN model optimized using the FOX optimization algorithm to select the most suitable augmentation methods. The results show that selecting appropriate augmentation techniques, such as K-means Color Quantization, Horizontal Flip, Voronoi, Elastic Transformation, and Contrast Normalization, can significantly improve the accuracy of fish species recognition, reaching up to 98.75 percent during the training phase. The proposed model also demonstrates strong generalization capabilities with a validation accuracy of 96.90 percent, indicating minimal overfitting. Although the training process is computationally intensive, this approach has proven to be highly effective for applications that require high accuracy and strong generalization capabilities, thus contributing to a better understanding and management of marine ecosystems in support of sustainable fisheries practices..]]>
