This research delves into deep learning methodologies for analyzing various datasets of dog images. It explores the utilization of convolutional neural network (CNN) architecture, which undergoes training subsequent to preprocessing steps ensuring data compatibility. Enhancing classification accuracy involves employing transfer learning from established models like VGG16 or ResNet. Evaluation metrics encompass accuracy, precision, recall, and F1-score, with confusion matrices revealing the model's efficacy across diverse dog breeds. The findings underscore the effectiveness of deep learning in discerning dog photographs, with implications spanning domains, such as veterinary care and animal welfare.