Desirable Model Properties for Vision

Typical tasks required when using a solid model for vision purposes include

• direct pairing of model features and observed data features,
• direct estimation of object position and orientation, and
• prediction of the appearance of the object from any position.