This paper provides a comprehensive overview of uncertainty estimation in deep neural networks (DNNs), highlighting recent advances, challenges, and potential research directions. It discusses the sources and types of uncertainty, including data uncertainty (aleatoric) and model uncertainty (epistemic), and explores various methods for estimating and quantifying uncertainty in DNNs. The paper reviews different approaches such as Bayesian neural networks (BNNs), ensemble methods, test-time augmentation, and deterministic networks with explicit components for uncertainty modeling. It emphasizes the importance of uncertainty estimation for safe decision-making in high-risk applications and for handling data with high inhomogeneity and limited labeled data. The paper also discusses the calibration of DNNs to ensure reliable uncertainty estimates and presents an overview of commonly used evaluation datasets, benchmarks, and implementations. It highlights real-world applications in fields such as medical image analysis, robotics, and earth observation, and discusses the practical limitations of uncertainty quantification methods in mission- and safety-critical applications. The paper concludes with an outlook on future research directions in uncertainty estimation for DNNs.This paper provides a comprehensive overview of uncertainty estimation in deep neural networks (DNNs), highlighting recent advances, challenges, and potential research directions. It discusses the sources and types of uncertainty, including data uncertainty (aleatoric) and model uncertainty (epistemic), and explores various methods for estimating and quantifying uncertainty in DNNs. The paper reviews different approaches such as Bayesian neural networks (BNNs), ensemble methods, test-time augmentation, and deterministic networks with explicit components for uncertainty modeling. It emphasizes the importance of uncertainty estimation for safe decision-making in high-risk applications and for handling data with high inhomogeneity and limited labeled data. The paper also discusses the calibration of DNNs to ensure reliable uncertainty estimates and presents an overview of commonly used evaluation datasets, benchmarks, and implementations. It highlights real-world applications in fields such as medical image analysis, robotics, and earth observation, and discusses the practical limitations of uncertainty quantification methods in mission- and safety-critical applications. The paper concludes with an outlook on future research directions in uncertainty estimation for DNNs.