Handwritten digit recognition is a fundamental problem in image processing with widespread applications in Optical Character Recognition (OCR), automated form processing, and document digitisation. This paper explores the implementation of a deep learning-based approach for handwritten digit recognition, emphasising the effectiveness of Convolutional Neural Networks (CNNs). The CNNs have proven to be highly efficient in extracting spatial features from images, making them well-suited for digit classification tasks. The proposed model is trained and evaluated on a custom dataset, with a structured preprocessing pipeline that includes normalisation, scaling and grayscale conversion to enhance input quality and improve model accuracy. The study highlights the significance of deep learning in image processing and demonstrates how CNN architectures contribute to robust and efficient digit recognition. The performance of the implemented model is assessed using standard evaluation metrics, and experimental results are presented to validate its effectiveness. The findings underscore the potential of CNN-based approaches in advancing automated handwritten digit classification and suggest avenues for further optimisation.