The usage of smartphones has rapidly increased during the last years. In addition to communication capabilities, they are also equipped with several sensors, and are usually carried by people throughout the day. The data collected by the means of modern smartphones (e.g. location based, GSM, and other contextual data) are thus valuable source of information for transportation analysis. In this paper we focus on smartphone data used for transportation mode detection. This is important for many applications including urban planning, context related advertisements or supply planning by public transportation entities. We present a review of the existing approaches for transportation mode detection, and compare them in terms of (i) the type and the number of used input data, (ii) the considered transportation mode categories and (iii) the algorithm used for the classification task. We consider these aspects as the most relevant when evaluating the performance of the analyzed approaches. Finally, the paper identifies the gaps in the field and determines future research directions.