2014 | OriginalPaper | Buchkapitel
Efficiency Analysis of Object Position and Orientation Detection Algorithms
verfasst von : Tomas Uktveris
Erschienen in: Information and Software Technologies
Verlag: Springer International Publishing
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
This work presents a performance evaluation of the state-of-the-art computer vision algorithms for object detection and pose estimation. Depth information from Kinect sensor is used in this work for the estimation task. It is shown, that Kinect depth sensor is more superior for orientation estimation than a regular stereo camera setup. Accuracy and performance of a point cloud alignment ICP method is analyzed and tested. Furthermore, multiple object detectors accuracy and runtime performance is evaluated. Simple but effective techniques are provided for the comparison. Conducted experiments show a maximum object detection accuracy of 90% and speed of 15 fps for standard size VGA images, while ICP alignment performance of 2 fps is achieved. Additional optimizations would be necessary to attain better real-time object detection and pose extraction performance.