A Dual-hierarchy Semantic Graph for Robust Object Recognition
We present a system for object recognition based on a semantic model graph, which it can learn automatically from image examples. This model graph is based on intrinsic properties of objects such as structure and geometry, so it is more robust than the current machine learning methods that can be fooled by changing a few pixels. Current methods have proved to be powerful but fragile because they ignore the structure and semantics of the objects. We define semantics, or abstraction, in terms of the intrinsic properties of the object, not in terms of human language, so it can be learned automatically. Our model graph is more versatile than previous ones because it uses two distinct hierarchies: parts and abstraction. Previous semantic networks used only one amorphous hierarchy and were hard to build and traverse. Our system performs both the learning and recognition by an algorithm that moves in both hierarchies at the some time, combining the advantages of top-down and bottom-up strategies. This reduces dimensionality and obviates the need for the brute force of big data training.
READ FULL TEXT