Publication: Development of face sketch to image matching using embedded system
Date
2024-07
Authors
Mohamed Hafzan Imtiaz
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In this study, an embedded system called the Raspberry Pi 4 is used to integrate a Cycle-Consistent Generative Adversarial Network (CycleGAN) model for the
development of face sketch synthesis. The major flaws in conventional facial recognition algorithms are addressed by this method, which frequently fails to generate realistic and personalized face sketches. This research attempts to produce precise and unique facial drawings from real-world photos by utilizing the capabilities of deep learning and embedded systems, improving applications in digital art and security. The research is initialised with dataset preparation, followed by the selection and setup of the Raspberry Pi 4 embedded system, and the assignment of training parameters for the CycleGAN model. Qualitative and quantitative analyses of the CycleGAN model's performance are conducted using metrics such as Structural Similarity Index Measure (SSIM), Feature Similarity Index Measure (FSIM) and Mean Squared Error (MSE). The findings demonstrate that the CycleGAN model, when trained on a diverse dataset, can generate high-quality face sketches that retain crucial facial features, making it suitable for practical applications in law enforcement and artistic projects. This study shows the feasibility of integrating advanced neural network models into compact and portable embedded systems but also lays the groundwork for future research and real-world implementations aimed at improving the face sketch creation.