Evaluation on the Compactness of Supervoxels.

Ran Yi,Yong-Jin Liu,Yu-Kun Lai

DOI: https://doi.org/10.1109/icip.2018.8451659

2018-01-01

Abstract:Supervoxels are perceptually meaningful atomic spatio-temporal regions in videos, which has great potential to reduce the computational complexity of downstream video applications. Many methods have been proposed for generating supervoxels. To effectively evaluate these methods, a novel supervoxel library and benchmark called LIBSVX with seven collected metrics was recently established. In this paper, we propose a new compactness metric which measures the shape regularity of supervoxels and is served as a necessary complement to the existing metrics. To demonstrate its necessity, we first explore the relations between the new metric and existing ones. Correlation analysis shows that the new metric has a weak correlation with (i.e., nearly independent of) existing metrics, and so reflects a new characteristic of supervoxel quality. Second, we investigate two real-world video applications. Experimental results show that the new metric can effectively predict some important application performance, while most existing metrics cannot do so.