Partial Occlusion in Facial Expression Recognition: A Systematic Literature Review

This study presents a Systematic Literature Review (SLR) examining Facial Expression Recognition (FER) under partial occlusion conditions. Analyzing a diverse array of studies, it identifies prevalent methodologies and challenges within the field. The review reveals a significant use of Gabor filters for feature extraction and classification, highlighting their efficiency in FER research. Additionally, black patches emerge as the primary method for simulating occlusions, which reflects the scarcity of datasets containing such occlusions. The complexity of FER under occlusion conditions is highlighted by the inconclusive findings on the relative importance of facial regions like the mouth and eyes in expression recognition. The results point to the need for more research to comprehend the complex role of different facial regions in conveying emotions. Furthermore, the review emphasizes the importance of developing larger and more diverse datasets, that encompass a broader range of facial expressions and occlusion types, to advance the development and evaluation of FER systems in real-world scenarios.