On using AI-Based Human Identification in Improving Surveillance System Efficiency
Eman Alajrami, Hani Tabash, Yassir Singer, Mohammed Taha O. El Astal · 2019
Surveillance camera systems are one of the most popular tech tools keeping homes and business safe. Many of these systems record continuously while some others start recording when a movement is detected. These recording methods consume a huge storage, and they need a dramatic time to search in the recorded files. Here, an AI-based desktop application which is designed and developed in order to start recording only if a person or a human-face is identified. This technique will improve system's efficiency in terms of reducing the storage needed for saving recordings, and reducing the processing and searching time in the recordings. This proposed system uses deep-learning algorithms, OpenCV libraries and built on Linux OS. It has two detection modes; human body identification and human-face identification. This will trigger the suitable detection algorithms based on mode selected. The system has been tested and evaluated by different evaluators and organizations that use surveillance systems, and questionnaire and interviews were conducted. Based on the evaluation results, it can be deduced how much this system is an important application to the target group.