An experiment using the Haar-Cascade and LBPH algorithms for real-time recognition of multiple faces in a single frame.

Face recognition is an important field in digital image processing and artificial intelligence that is widely applied in security systems, automatic attendance, and human-computer interaction. This research aims to develop and test a real-time multiple face recognition system using a combination of the Haar Cascade algorithm for face detection and the Local Binary Pattern Histogram (LBPH) for face recognition. The system is implemented using the Python programming language and the OpenCV library, and tested under various conditions, such as variations in lighting, face viewing angles, and the number of faces in one frame. Test results show that the system is able to recognize multiple faces with approximately 90% accuracy under normal lighting conditions with varying distances, and maintains performance above 80% under low lighting conditions or side-facing face angles. The average detection and recognition time per face ranges from 40–60 milliseconds, which still supports real-time performance. Compared with deep learning-based approaches, this system has advantages in terms of efficiency and ease of implementation, especially on devices with limited specifications. This study shows that the combination of Haar Cascade and LBPH is still relevant and effective for light- to medium-scale multiple face recognition applications. Keywords: Multiple faces, Haar Cascade, Face recognition, LBPH