The human face contains important and understandable information such as personal identity, gender, age, and ethnicity. In recent years, a person's age has been studied as one of the important features of the face. The age estimation system consists of a combination of two modules, the presentation of the face image and the extraction of age characteristics, and then the detection of the exact age or age group based on these characteristics. So far, various algorithms have been presented for age estimation, each of which has advantages and disadvantages. In this work, we implemented a deep residual neural network on the UTKFace data set. We validated our implementation by comparing it with the state-of-the-art implementations of different age estimation algorithms and the results show 28.3% improvement in MAE as one of the critical error validation metrics compared to the recent works and also 71.39% MAE improvements compared to the implemented AlexNet. In the end, we show that the performance degradation of our implemented network is lower than 1.5% when injecting 15 dB noise to the input data (5 times more than the normal environmental noise) which justifies the noise tolerance of our proposed method.