HOG Descriptors can be uesd to identify an object with a sliding window. It is very slow, so the goal is to train an SVM on the data to speed up the process. This method is much more reliable then the color segmentation we had previously relied on. Below is a snapshot of the selection window in action:
