Makine öğrenimi görsel akış sistemlerinde gizlilik ve güvenliği artırma

This thesis delves into the challenges and opportunities presented by the integration of new security layers into machine learning (ML) visual stream systems, primarily used in surveillance and security. The study is anchored on the premise that while the advancement of ML in visual processing offers significant benefits, it also raises critical concerns regarding data privacy and system vulnerabilities. A key focus of this research is the implementation of additional security layers into the existing ML system architecture. This approach is examined not only in terms of enhancing data protection and system resilience but also in assessing its impact on performance metrics. The research aims to quantify and analyze the trade-offs between system security and performance, providing valuable insights into the development of ML visual stream systems that are both secure and efficient. The findings are expected to contribute to a more balanced approach in designing ML systems that are robust against evolving digital threats while maintaining optimal performance.