Facial recognition technology has become an essential tool for various security applications. However, traditional face recognition systems require several images of an individual to be effective. Training systems with only a single image per person remains a significant challenge. This paper presents a face-recognition approach using a single image by leveraging deep learning techniques. We propose a method that generates multiple variations of a single image through image segmentation and preprocessing techniques, such as de-noising and illumination correction. The core of our approach involves using Siamese neural networks, which are trained to distinguish between similar and dissimilar image pairs, enabling accurate recognition even with minimal input data. Our methodology includes developing and training a Siamese neural network focusing on metric-based learning for one-shot face recognition. We evaluate our approach on the Celebrity dataset, achieving 97.5% accuracy and demonstrating competitive performance compared to existing single-image face recognition methods, even with limited input data.