• Dr.Sanju S Anand
• Rathish
• Franklin Roy Pereira
• Sushanth
• Kalmesh M Jogin

Face recognition has evolved significantly with deep convolutional neural networks, large-scale face datasets, discriminative embedding-learning techniques, and efficient face detection architectures. This research presents the design, development, and deployment of a complete deep-learning-based face recognition system using Python for model preparation and reference embedding generation and C#/.NET for the final desktop application.The proposed system integrates the Sample and Computation Redistribution for Efficient Face Detection (SCRFD) model for face localization and five-point facial landmark extraction with a ResNet50-based face recognition model associated with the WebFace600K training framework. The models are used in ONNX format and executed in the C# application using Microsoft ONNX Runtime. OpenCvSharp is employed for image processing, resizing, padding, facial alignment, and pixel manipulation.The system follows a multi-stage pipeline consisting of image acquisition, face detection, facial landmark localization, facial alignment, 112 × 112 face normalization, deep feature extraction, 512-dimensional embedding generation, L2 normalization, cosine similarity calculation, and identity matching against a local embedding database. Python is used during model preparation and enrollment, while final inference operates within a C# Windows Forms environment without requiring a Python runtime.The prototype successfully detected faces and extracted five facial landmarks, with an observed SCRFD detection confidence of 0.533 for one test image. The complete recognition pipeline generated facial embeddings and performed identity matching against stored reference embeddings. A systematic evaluation framework is provided for recognition accuracy, precision, recall, F1-score, false acceptance rate, false rejection rate, similarity distributions, and inference time.

"An AI-Based Deep Learning Framework for Face Recognition Using SCRFD and ResNet50 with ONNX Runtime", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 9, page no.c579-c594, September-2026, Available :http://www.jetir.org/papers/JETIR2609264.pdf

"An AI-Based Deep Learning Framework for Face Recognition Using SCRFD and ResNet50 with ONNX Runtime", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 9, page no. ppc579-c594, September-2026, Available at : http://www.jetir.org/papers/JETIR2609264.pdf