Face Detection in Security Monitoring Based on Artificial Intelligence Video Retrieval Technology

• Published in: IEEE access Vol. 8; pp. 63421 - 63433
• Main Authors: Dong, Zuolin, Wei, Jiahong, Chen, Xiaoyu, Zheng, Pengfei
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial intelligence; Computer vision; Deep learning; Face; Face detection; Face recognition; Feature extraction; Human resources; Image processing; Learning theory; Machine learning; Monitoring; Monitoring systems; Neural networks; New technology; Retrieval; Security; security monitoring; Semantics; Streaming media; Video; video retrieval
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2020.2982779

With the rapid development of video monitoring, the massive information of the monitoring image has far exceeded the effective processing range of human resources. Intelligent video retrieval technology has become an increasingly indispensable part of video monitoring system. Intelligent video retrieval technology integrates video processing, computer vision and artificial intelligence, which greatly improves the efficiency of monitoring and the accuracy and linkage of monitoring system. Face recognition and other emerging technologies continue to rise and apply to the security monitoring system. Based on deep learning theory and face detection neural network, this paper proposes a video oriented cascaded intelligent face detection algorithm, which builds deep learning network by cascading multiple features, from edge features, contour features, local features to semantic features, and advances layer by layer. According to the last semantic features, the information of the input data is obtained to accurately realize the face detection under the non ideal condition. Simulation results show that the algorithm has good detection performance for single face and multi face images, and has strong robustness for rotating face. At the same time, the algorithm is fast and can basically meet the requirements of real-time face detection.