Feature Harvesting for Tracking-by-DetectionReport as inadecuate

Feature Harvesting for Tracking-by-Detection - Download this document for free, or read online. Document in PDF available to download.

Presented at: European Conference on Computer Vision, Graz, May 7-13, 2006 Published in: Computer Vision - ECCV 2006, p. 592-605 Series: Lecture Notes in Computer Science 3953 Berlin / Heildelberg: Springer, 2006

We propose a fast approach to 3-D object detection and pose estimation that owes its robustness to a training phase during which the target object slowly moves with respect to the camera. No additional information is provided to the system, save a very rough initialization in the first frame of the training sequence. It can be used both to detect the target object in each frame independently and to track it from frame to frame. Our approach relies on a Randomized Tree-based approach to feature-point wide-baseline matching. Unlike previous classification-based approaches to 3-D pose estimation, we do not require an a priori 3-D model. Instead, our algorithm learns both geometry and appearance. In the process, it collects, harvests, a list of object features that can be reliably recognized even when large motions and aspect changes cause complex variations of feature appearances. This is made possible by the great flexibility of Randomized Trees, which lets us add and remove feature points to our list as needed with a minimum amount of extra computation.

Keywords: Computer Vision ; Object Detection ; Tracking by Detection ; Pose Estimation Reference CVLAB-CONF-2008-003

Author: Ozuysal, Mustafa; Lepetit, Vincent; Fleuret, Francois; Fua, Pascal

Source: https://infoscience.epfl.ch/record/126375?ln=en

Related documents