After reading the paper, I have no idea whether they mean classifying "vehicle" vs "non-vehicle", or "2013 Ford Mustang".

Based on the thesis [1] of one of the authors and the dataset [2] it links to, I believe it's {bus,truck,van,car} classification on a dataset of only vehicles.

[1]: Dataflow-Based Implementation of Deep Learning Application https://dspace.cc.tut.fi/dpub/bitstream/handle/123456789/240...

[2]: Car Type Recognition with Deep Neural Networks https://arxiv.org/pdf/1602.07125v2.pdf