Target detection is a research topic which is widely used in many fields. The traditional methods of computer vision have some problems, such as high complexity, poor generalization ability and weak mobility of model objects. However, human brain has higher cognitive ability and perception ability in object detection, only a few training samples are needed to achieve better detection results. The Brain-Computer Interface (BCI) based on Event-Related Potentials (ERP) is a new method to improve the performance of target detection, which can interact with information in the brain and the external environment. On this basis, a new pattern of Rapid Serial Visual Presentation (RSVP) based on EEG signals is proposed by Gerson, it includes display mode, effective visual range, target complexity, stimulus frequency and duration. In addition, the methods of ERP preprocessing, feature extraction and classification are reviewed. Finally, the limitation and development trend of this field are discussed.