Sexing Orcas: Classification Using Convolution Neural Networks

Identifying individual megafauna, such as orcas, is essential for understanding population dynamics, migration, and behavior. However, traditional methods like manual photo identification are time-consuming and resource-intensive. This study leverages artificial intelligence to automate orca sex classification, a crucial first step toward the unique identification of individuals. A convolutional neural network model, enhanced through transfer learning with VGG16, was developed to classify orcas as male or female. Using datasets from organizations such as the Norwegian Orca Survey and the Center for Whale Research, the model achieved accuracy rates of 93% on training, 94% on validation, and 95% on test sets. This research demonstrates the potential of AI to revolutionize wildlife monitoring by improving efficiency and scalability. While challenges such as image quality and overfitting persist, the adaptability of the approach offers promising applications for conservation across diverse species.