Vehicle detection in video is an important problem in Computer Vision because of the potential applications in security, vehicle traffic, driving assistance and so on. In this work, we used Mixture of Deformable Part Models (MDPM) for vehicle detection in video sequences obtained from static and dynamic cameras. The MDPM method was originally proposed by Felzenszwalb et al in the realm of object detection in images. We tested this method in the realm of video sequences for vehicle detection. We designed a set of experiments that explore the number of components of the mixture and the number of parts model. We performed a comparison study of symmetric and asymmetric MDPMs for vehicle detection. Our findings show that not only the MDPM performed well in vehicle detection in video, but also the best number of components and parts model confirmed the number suggested in Felzenzwalb et al's paper. Finally, the results show some differences between the symmetric and asymmetric MDPMs in vehicle video detection considering different scenarios.