This study aims to provide a comparative analysis of three machine learning algorithms: Convolutional Neural Networks (CNNs), Fully Connected Neural Networks, and Random Forests. These algorithms are trained and tested against the CIFAR-10 dataset. Which consists of 60,000 images across 10 object classes. We evaluate the performance and efficiency of different architectures when handling image classification tasks. The research systematically compares each model's ability to generalise and learn patterns hidden within images, with focuses on their performance efficiency and architectural strengths. The study provides insights into the practical trade-offs between model complexity and classification accuracy.