Human age recognition holds considerable importance in various social and medical contexts. Facial evaluation can be challenging due to variables such as photography equipment, lighting, and camera angles. Many image-based aging applications can take advantage of facial texture features, which are typical features of aging human skin. Age prediction can be viewed as either a regression or classification problem from a research methodology standpoint. This paper explores a method for approximating age based on facial texture attributes. The textural representation of facial features includes details like wrinkles around the eyes and cheeks. The proposed framework involves pre-processing where the frontal face is extracted and resized, texture feature extraction, and age group estimation using CNN. In this study, Local Binary Pattern (LBP), Gabor, Local Phrase Quantization (LPQ), Wavelet transform, Histogram of Oriented Gradients (HOG), Binarized Statistical Image Feature (BSIF), are utilized for feature extraction and early_stop as model regularization parameter to stop model overfitting during training. The model was trained using the publicly available UTKFace dataset and tested on our dataset consisting of 3017 face images in the range of 0–95 years. From the experiments, we observed that LPQ and DWT outperform other texture feature methods with a testing accuracy of 77% and 75% with regularization. The results show that the texture-based feature effectively determines the facial age group from the image size of 48 × 48, and model regularization techniques enhance the performance.