ANOMALY DETECTION SMART CAMERA

Abstract

This report introduces a technique to identify default camera events using image analysis. The key feature of our project is to ensure good image quality and to provide appropriate platform for monitoring surveillance videos. The approach of our project is to remove the reduced referenced features in most regions of the surveillance image and then to detect anomaly related scenarios by studying the variation of features when the viewing field changes. Real-time alerts for video surveillance systems have made anomaly camera detection increasingly important. However, existing methods are inadequate in detecting various abnormalities and are incapable of self-study to improve their performance in case of failures. This report proposes Anomaly camera detection method that uses morphological analysis and in-depth reading to detect a wide range of anomalies. Morphological analysis is used for detecting simple anomaly cameras detection to improve processing speed, while in-depth reading is used for identifying complex anomaly camera distractions to enhance accuracy. The proposed technique has been tested and proven to have an accuracy rate of over 95% results are further elaborated in the report. Our project starts with the previous process of video capture with limited video, inverting, sliding, sorting, resizing, and extracting features are also done in this process. Next, a network feed process is requested to generate an output matrix. Based on the output matrix, the known Face, object or character can be determined. This project is designed to customize the network to each user. Recommendations for future development and conclusions are also included in the report