Face Expression Recognition (FER) has become challenging area in computer vision research due to its wide range of applications. The FER includes comprehending human behaviour, recognizing mental states and facilitating human-computer interaction. FER has become an important field in facial image processing due to its adaptability. The numerous applications of FER highlight the development of facial image processing systems. The FER aids in the comprehension of non-verbal clues, offering perceptions into people’s intentions and attitudes in domains like human behaviour understanding. Numerous datasets with various features have been created for the purpose of training and assessing FER systems. This study discusses the datasets that the researchers frequently use, their characteristics and contents and how the data is generated. Additionally, gathering innovative technologies and diverse approaches that are used to datasets in order to achieve the best possible accuracy rate is the main feature of this review. The adaptability and impact of FER highlight the importance of developing advanced facial image processing systems. This review explores the challenges, datasets, and technologies associated with FER, providing a comprehensive overview of the current state and future directions in this evolving field. Every dataset offers insights on distinctive characteristics such subject demographic variety, feelings and annotation techniques. It provides information about the types of datasets that are currently accessible and suggests future paths for the development, application and assessment of datasets in the exciting field of facial expression recognition.