This paper presents a containing ‘ground truth’ segmentations produced by humans for images of a wide variety of natural scenes-This paper presents a containing ‘ground truth’ segmentations produced by humans for images of a wide variety of

The uation of performance of edge detectors in general, is made empirically by visual analysis of the detected edge map. This paper presents a new method that alow us to uate quantitatively the performance of edge detectors, through a set of

We propose a learning-based approach for motion boundary detection. Precise localization of motion boundaries is essential for the success of optical fl ow estimation, as motion boundaries correspond to discontinuities of the optical fl ow

基于类似活动的聚类，提出了一种用于生成简短而一致的视频摘要的新方法。 具有相似活动的物体可以同时观看，并可立即发现异常值。 群集概要也适用于有效创造地面真相数据。-A new methodology for the generation of short and coherent video summaries is presented, based on clustering of similar activities. Objects with similar activities are