One of the most crucial uses for facial analysis is facial diagnosis. Practitioners of traditional medicine have used their expertise to assess a person's health state for tens of thousands of years. Since then, a lot has changed, and this long-standing tradition has been supplanted by computerised face diagnosis. The proposed method of utilizing deep transfer learning from face recognition for face illness diagnosis with computer assistance is an interesting approach that can potentially provide a low-cost and non-invasive way for disease screening and detection. The objective of this research endeavor is to investigate whether techniques based on deep learning may be employed to detect malignancies through uncontrolled 2D face images. By leveraging the pre-trained face recognition models, the system can effectively learn relevant features from the face images and use them to diagnose different diseases. The proposed system will be using data augmentation to handle the imbalance of data in the system and reduce over-fitting. By artificially generating new data from the existing dataset, the system can increase the amount of data and diversity of the samples, leading to better generalization and improved accuracies.