An object recognition system finds objects in the real world from an image of the world, using object models which are known. This task is surprisingly difficult. Humans perform object recognition effortlessly and instantaneously. Algorithmic description of this task for implementation on machines has been very difficult.
- steps in object recognition
- techniques that have been used for object recognition
- types of recognition tasks that a vision system may need to perform
- complexity of these tasks
- Traditional Method – This method is appropriate when the object you want to recognize follows readily identifiable patterns, attributes and rules. These characteristics make it easier to distinguish among different types of objects. For example, such attributes can be the size, shape and pattern of the tread on a particular type of car tire.
- Machine Learning Method – This method is appropriate when the objects of interest share common characteristics but must be distinguished and a huge number of variations must be taken into account. To use a machine learning solution, a large number of images is required for learning to occur.
- Blended Methods – When both methods are needed, a blended approach is best.
Computers are getting better at object recognition
MegaFace Challenge – forces facial recognition algorithms to do verification and identification, two separate but related tasks. Verification involves trying to correctly determine whether two faces presented to the facial recognition algorithm belong to the same person. Identification involves trying to find a matching photo of the same person among a million “distractor” faces.
What’s the problem?
Face Recognition (video 2:43) for identification verification.
- mimic, visual, recognition, neuroscience, model, neural network, process, graphical processing unit (GPU), algorithms, dataset, annotation, computational tools