Video Anomaly Event Detection (VAED) is the core technology of intelligent surveillance systems aiming to temporally or spatially locate anomalous events in videos. With the penetration of deep learning, the recent advances in VAED have diverged various routes and achieved significant success. However, most existing reviews focus on traditional and unsupervised VAED methods, lacking attention to emerging weakly-supervised and fully-unsupervised routes. Therefore, this review extends the narrow VAED concept from unsupervised video anomaly detection to Generalized Video Anomaly Event Detection (GVAED), which provides a comprehensive survey that integrates recent works based on different assumptions and learning frameworks into an intuitive taxonomy and coordinates unsupervised, weakly-supervised, fully-unsupervised, and supervised VAED routes. To facilitate future researchers, this review collates and releases research resources such as datasets, available codes, programming tools, and literature. Moreover, this review quantitatively compares the model performance and analyzes the research challenges and possible trends for future work.