Spacetime Structure Grouping Dataset

Overview

This webpage provides datasets for spatiotemporal grouping. The dataset contains 11 image sequences that depict a wide range of juxtaposed spatiotemporal patterns, including motion, static (stationary texture), pseudo-transparency, translucency, temporal flicker, unstructured and various additional dynamic textures. The image sequences were acquired using a Canon HF10 consumer camcorder with capture rate of 30fps. Each image sequence is provided with a manually constructed groundtruth that delineates boundaries between distinct spatiotemporal regions.

For details of our technical approach to spacetime structure grouping, see our project page.

Examples

Input	Ground Truth
Example 1: Tree in foreground (static structure) being coarsely stabilized by moving camera operator with resulting background motion (motion structure).
Example 2: A painting (static structure) hanging on a blank wall (unstructured) with a light flickering in an adjacent hallway (temporal flicker structure).
Example 3: A duck swimming (motion structure) over turbulent water (dynamic texture structure).
Example 4: A person walking behind a fence (pseudo-transparency structure) with a stationary surround of fence, grass and trees (static structure).

Download

• YUVL-STS (York University Vision Lab - SpaceTime Structure) Grouping Dataset (27 megs)


• Supplementary video of grouping results for CVPR2009 (requires DIVX, 19 megs)

Related Papers

• K.G. Derpanis and R.P. Wildes, Detecting Spatiotemporal Structure Boundaries: Beyond Motion Discontinuities, In Proceedings of the Asian Conference on Computer Vision, to appear, 2009.


• K.G. Derpanis and R.P. Wildes, Early spatiotemporal grouping with a distributed oriented energy representation, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2009.


• R.P. Wildes and J.R. Bergen, Qualitative spatiotemporal analysis using an oriented energy representation, European Conference on Computer Vision (ECCV), pp. 786-784, 2000.