A large portion of the money is utilized every year to find defects in important infrastructure, such as roads, bridges, and buildings. The urban infrastructure is severely damaged by natural calamities like earthquakes and floods. The ensuing maintenance procedures typically entail a visual examination and appraisal of the compromised infrastructure to guarantee its structural and operational soundness. This damage may initially manifest as tiny or large fractures, which may ultimately lead to the building’s complete collapse. The method of manually identifying cracks through visual inspection is somewhat time-consuming. It is not feasible since it will need many human resources to verify each component of the system on a regular basis. Furthermore, this may lead to situations in which fissures remain hidden. Therefore, to guarantee the efficacy and dependability of infrastructure, automatic fault detection must be carried out. Image processing techniques can be used on scanned or taken photographs of the infrastructure components to find any potential flaws. To improve performance results and reliability in crack detection, machine learning techniques are being used more and more in addition to image processing. An overview of image-based crack detection methods utilizing machine learning or image processing is given in this paper.