Deep Learning for Classification of Mammalian Reproduction

Abstract

This research aims to classify mammalian reproduction using deep learning techniques, specifically focusing on Convolutional Neural Networks (CNN), VGG16, and MobileNetV2. CNN is applied to extract visual features from images related to mammalian reproductive systems, while the VGG16 and MobileNetV2 architectures are utilized to enhance accuracy and efficiency in classification. The dataset used consists of images of mammalian reproductive organs, analyzed using data augmentation techniques to improve model reliability. Data augmentation includes various transformations such as rotation, zoom, flipping, and brightness adjustments to enrich the dataset's variety and reduce the risk of overfitting. The results of this study indicate that the combination of these methods achieves high accuracy. VGG16 demonstrates the best performance in terms of precision, achieving an accuracy of 90.97%. MobileNetV2, while slightly less accurate (65.97%), excels in computational efficiency, making it highly suitable for mobile and resource-constrained environments. The baseline CNN model, achieving an accuracy of 61.11%, shows that simpler architectures are less effective in handling the complexity of the dataset. The implementation of this technology is expected to support more accurate and automated analysis and diagnosis in mammalian reproduction. The findings of this research provide valuable insights into the strengths and weaknesses of each architecture, as well as the trade-offs between accuracy and computational efficiency. The study also highlights the importance of using data augmentation techniques to improve the quality and diversity of the dataset, which in turn enhances model performance.