detect object in image,how they did it. I think it

问题:

alt text http://pascallin.ecs.soton.ac.uk/challenges/VOC/voc2007/examples/person_06.jpg alt text http://pascallin.ecs.soton.ac.uk/challenges/VOC/voc2007/examples/dog_08.jpg alt text http://pascallin.ecs.soton.ac.uk/challenges/VOC/voc2007/examples/sheep_07.jpg

this picture is download from a website,the team show their research result like this. I want to know how they did it, some body know tell me I will be appreciate. thank you

回答1:

Based on their publications, one can conclude they use SVM classifier and/or some sort of a graphical model (Markov Random Field / Conditional Random Field). Also, they (probably) use Histograms of Oriented Gradients as features. The approach implies really complicated math (machine learning, to be precise). If you are not familiar, you could try to use their implementation.

回答2:

The algorithm was published.

And you can redo it yourself, if you do have your Math skills sharp enough.