Fast object detection using steiner tree

• Authors: Duanmu Xiaoyin
• Venue: Machine Graphics & Vision International Journal
• Year: 2010

We propose an approach to speed-up object detection, with an emphasis on settings where multiple object classes are detected. Our method uses a segmentation algorithm to select a small number of image regions on which to run a classifier. Compared to the classical sliding window approach, a significantly smaller number of rectangles is examined, which yields significantly faster object detection. Further, in the multiple object class setting, we show that the computational cost of segmentations can be amortized across objects classes, resulting in an additional speedup. At the heart of our approach is reduction to a directed Steiner tree optimization problem, which we solve approximately in order to select the segmentation algorithm parameters. The solution gives a small set of segmentation strategies that can be shared across object classes. Compared to the sliding window approach, our method results in two orders of magnitude fewer regions considered, and significant (10-15x) computational time speedups on challenging object detection datasets (LabelMe and StreetScenes) while maintaining comparable detection accuracy.