Concrete crack detection and classification are pivotal for structural health monitoring and civil infrastructure maintenance. However, traditional manual inspections are labour-intensive, inconsistent, and often hazardous in complex environments. Conventional manual inspections are time-consuming and could turn out ineffective, leading to risks in challenging circumstances. To address these limitations, a systematic literature review (SLR) of 68 peer-reviewed studies published between 2021 and 2025 was conducted, examining the evolution of concrete crack detection and classification methods over the past decade. This systematic review organises recent concrete crack detection and classification studies based on the crack analysis pipeline (data acquisition, preprocessing, model development, and performance evaluation) and identifies the key techniques employed at each stage. This review evaluates the accuracy, practical usefulness, limitations, and implementation challenges of recent concrete crack detection and classification methods, as well as their practical problems and ongoing problems, such as limited training datasets, variable environmental conditions, real-time deployment constraints, and limited model interpretability. To fill these gaps, the review suggests future research directions like creating standard crack datasets and making lightweight, easy-to-understand AI models that can be directly implemented in field applications. The review shows a clear shift from traditional image processing and classical machine learning methods to more advanced deep-learning models, which have improved automated crack detection capability. Convolutional neural networks and transformer-based architectures have reported strong performance in selected studies, although these values should be interpreted within the context of different datasets, validation protocols, and evaluation metrics. This greatly improves the safety and efficiency of inspections. Consequently, deep learning has emerged as one of the most widely adopted and promising approaches for concrete crack detection and classification during the reviewed period.