Image classification is one of the basic primary keys in computer vision tasks. As classification techniques continue to evolve, classifiers need to accurately classify into multiple classes. This study examines the performance of Residual Network (ResNet) for multi-class image classification using an augmented dataset of 2,500 images across six categories: adult, male, female, child, boy, and girl. The model was designed to assign multiple class labels to a single image, which may contain one or more objects, thereby facilitating multi-class classification. The ResNet-18 model was trained for 50 epochs and achieved an accuracy of 0.82 for multi-class predictions. However, the model faced challenges in certain scenarios due to limitations within the dataset. The evaluation demonstrated that the classifier performed reasonably well, but it struggled with ambiguous cases, resulting in errors. The ResNet-18 model developed in this study can be effectively combined with smaller or more compact object detection algorithms to classify detected objects into multiple classes without affecting the real-time performance.