Gender identification of face images is one of the most important applications of computer vision for use in security, demographic monitoring, and human–computer interaction. With the advent of deep learning, convolutional neural networks (CNNs) are able to obtain state-of-the-art results in this application. This paper performs a comparative investigation of two popular deep CNN architectures, ResNet50 and ResNet101, for the identification of gender. Both of the networks have been trained and tested on a balanced dataset of female and male face images with a standard preprocessing pipeline of face detection, normalization, and data augmentation. The performance of the algorithms was evaluated based on accuracy, precision, recall, and F1-score. Findings indicate that even though ResNet101 had marginally better classification accuracy, ResNet50 had quicker inference and reduced computational complexity and hence could serve better for real-time application, ResNet101 gained slightly higher classification accuracy of 94.80%, whereas the accuracy of 93.60 was produced by ResNet50. These results indicate the associated trade-offs of model depth, accuracy of prediction, and efficiency and have useful implications for deep learning-based gender classification on resource-poor platforms.