Gender classification has become relevant to applications since the rise of social platforms and social media. The performance of existing methods of real-world images is still eloquently lacking, especially as compared to the top-notch leaps in overall performance. Gender classification of human faces is studied by introducing a structure that includes deep learning models (DL) like convolutional neural networks (CNN), which helps dealing with problems related to gender classification. Three models are proposed, which include three hybrid models, viz., VGG16 + CNN, CNN + LSTM and CNN, where it was found that the CNN performance was significantly better than the other two models. CNN is a form of deep learning model, which is represented for large-scale images of gender classification and related works. The dataset used in this research for training and testing consisted of more than two hundred thousand images, on which an accuracy of 97% was obtained.