Road surface condition is vitally important for road safety and transportation efficiency. Conventionally, road surface monitoring relies on specialised vehicles equipped with professional devices, but such dedicated large-scale road surveying is usually costly, time-consuming, and prohibitively difficult for frequent pavement condition monitoring—for example, on an hourly or daily basis. Current advances in technologies such as smartphones, machine learning, big data, and cloud analytics have enabled the collection and analysis of a great amount of field data from numerous users (e.g., drivers) whilst driving on roads. In this regard, we envisage that a smartphone equipped with an accelerometer and GPS sensors could be used to collect road surface condition information much more frequently than specialised equipment. In this study, accelerometer data were collected at low rate from a smartphone via an Android-based application over multiple test-runs on a local road in Ireland. These data were successfully processed using power spectral density analysis, and defects were later identified using a k-means unsupervised machine learning algorithm, resulting in an average accuracy of 84%. Results demonstrated the potential of collecting crowdsourced data from a large population of road users for road surface defect detection on a quasi-real-time basis. This frequent reporting on a daily/hourly basis can be used to inform the relevant stakeholders for timely road maintenance, aiming to ensure the road’s serviceability at a lower inspection and maintenance cost.