Motorbikes

On the left we show the shape model. Each part of the model is shown in a different colour. The mean location is indicated by the cross, with the ellipse showing the uncertainty in location. The number by each part is the probability of that part being present. On the right, we show examples of regions which are closest to the mean of the appearance density of each part. We can see that the wheels of the motorbike are cleanly picked up, along with the rear of the bike and the fuel tank and the handlebars/front fork region.

We now show the model picking out motorbikes in various test frames. The size of the ellipse corresponds to the size of the region:

Here is the model evaluating some background images. The best assignment is shown, although the majority fall below the threshold so are correctly classified.