Facial Image Recognition Using Transfer Learning: A Brief Review

Accurate image recognition plays a pivotal role in a wide range of industrial applications. One domain of these applications is facial image recognition. Recently, deep learning (DL) models, which provides high efficiency and accuracy, have been utilized in facial image recognition applications. However, DL-based solutions in such applications need huge amount of data for training the model which will result in increasing the learning duration. Recently, the emergence of transfer learning (TL) has become an innovative solution in enhancing and improving the learning process in developing the machine learning (ML) models for facial image recognition-based applications. optimization of such systems.TL is a kind of DL models which guarantee enhancements in the learning process where the TL-based models don’t need to start the training process from scratch. Moreover, TL ensures effective use of limited data by leveraging pre-trained models and start the learning process from a certain level. Therefore, TL developers can create accurate and powerful AI solutions faster and with fewer resources. This paper presents a brief review for the recently published works in the field of facial image recognition based on transfer learning. The review shows a description of the dataset used in the training process, methodology, and the performance results of each work. Moreover, it highlights the challenges of using TL in facial image recognition-based applications. Some future work directions are also presented in the paper.