In this work we present a mobile pill recognition system for conventional smartphones. An algorithm for segmentation of pills from a business card sized marker is presented. Hereafter, relevant properties, such as size, shape and color are chosen to be determined by means of CV. A domain analysis is carried out to identify the most relevant shapes and colors of pills. The estimated features can be used for identification of medical pills with in a database then. An example of this application interfacing an online available database for medical matters is presented, followed by an extensive evaluation of the system interms of algorithm speed and pill retrieval performance. The obtained results show that the approach is able to work in instant time using commonly available smartphone hardware. Furthermore, the retrieval performance of the system on the ex emplarily used Ident a database confirms that the system can facilitate the task of visual pill recognition in a realistic scenario. For the verification of practical use an extensive user study will be necessary. The incorporation of additional features such as brick lines or text could be the topic of future work.