Motorcyclist detection

Hello

I try to detect motorcyclists on a videostream and I'd like to try out these hot deep learning models. I've found a promising article from LeCun's lab (Pedestrian Detection with Unsupervised Multi-Stage Feature Learning: http://cs.nyu.edu/~sermanet/papers/sermanet-cvpr-13.pdf) and going to stick with it. But i'd also like to hear community opinions on my problem.

Do you know other approaches? Or do you have something to say about aforementioned paper? Thanks in advance. P.S. I'm not interested in computer vision's hand-engineered features since I want to play with deep learning