Abstract: Person Re-Identification (Re-ID) has been a problem recently faced in computer vision. Most of the existing methods focus on body features which are captured in the high-end surveillance system. However, it is unhelpful for authentication. Automatic face recognition for still images with high quality can achieve satisfactory performance, but for video-based face recognition it is hard to attain similar levels of performance. Compared to still images face recognition, there are several disadvantages of video sequences. First, images captured by CCTV cameras are generally of poor quality. The noise level is higher, and images may be blurred due to movement or the subject being out of focus. Second, image resolution is normally lower for video sequences. If the subject is very far from the camera, the actual face image resolution can be as low as 64 by 64 pixels. Last, face image variations, such as illumination, expression, pose, occlusion, and motion, are more serious in video sequences. In the face recognition approach for controlled scenario to authenticate a person. Initially, faces are detected by Generative Adversarial Neural Network (GAN) and landmark points are obtained using Supervised descent method (SDM) Finally, face is recognized by Joint Bayesian model and provide the voice, Sms and E-mail alert at the time of unknown face detection in real time environment. The proposed framework overcomes the challenges such as pose variations, low resolution and partial occlusion. The experimental results (accuracy) on benchmark dataset demonstrate the effectiveness of the proposed method.
Keywords: Video surveillance, Person re-identification, Face recognition and SDM.
| DOI: 10.17148/IJARCCE.2020.9521