The paper presents a generic objectness measure that quantifies the likelihood of an image window containing any object, distinguishing well-defined objects from amorphous backgrounds. The measure combines several image cues in a Bayesian framework, including multi-scale saliency (MS), color contrast (CC), edge density (ED), superpixels straddling (SS), and location and size (LS). These cues capture characteristics such as closed boundaries, different appearances, and uniqueness. The parameters of these cues are learned from a training dataset using a Bayesian approach. The combined objectness measure outperforms individual cues and traditional saliency measures on the challenging PASCAL VOC 07 dataset. The paper also demonstrates applications of objectness in reducing the number of windows evaluated by class-specific object detectors and in reducing false positives. The method is efficient, taking only about 4 seconds per image to compute.The paper presents a generic objectness measure that quantifies the likelihood of an image window containing any object, distinguishing well-defined objects from amorphous backgrounds. The measure combines several image cues in a Bayesian framework, including multi-scale saliency (MS), color contrast (CC), edge density (ED), superpixels straddling (SS), and location and size (LS). These cues capture characteristics such as closed boundaries, different appearances, and uniqueness. The parameters of these cues are learned from a training dataset using a Bayesian approach. The combined objectness measure outperforms individual cues and traditional saliency measures on the challenging PASCAL VOC 07 dataset. The paper also demonstrates applications of objectness in reducing the number of windows evaluated by class-specific object detectors and in reducing false positives. The method is efficient, taking only about 4 seconds per image to compute.