- Title
- A study on detecting three-dimensional balls using boosted classifiers
- Creator
- Metcalfe, Mitchell; Annable, Brendan; Olejniczak, Monica; Chalup, Stephan K.
- Relation
- ACSW '16 Australasian Computer Science Week Multiconference. ACSW '16 Proceedings of the Australasian Computer Science Week Multiconference (Canberra, A.C.T. 01-05 February, 2016)
- Publisher Link
- http://dx.doi.org/10.1145/2843043.2843473
- Publisher
- Association for Computing Machinery (ACM)
- Resource Type
- conference paper
- Date
- 2016
- Description
- Many recent approaches to ball detection in robot soccer reduce the task to edge-based circle detection, or training a classifier to detect specific balls with known colour or surface texture. In the present work, a more general approach to ball detection is investigated, where spherical 3D objects must be detected under unknown lighting, colouring and texturing. Pilot experiments applied techniques stemming from the face detection literature, namely boosted-classifiers using extended Haar features, and Local Binary Patterns (LBPs) as features. Disk-like objects were included as negative samples in the training set in order to produce a detector that does not misclassify circular, disk-like objects as 3-dimensional balls. The resulting classifiers were able to detect homogeneously or moderately textured balls while robust detection of balls with unknown strong patterns still remains a challenge.
- Subject
- computer vision; sphere detection; adaboost; classification
- Identifier
- http://hdl.handle.net/1959.13/1343852
- Identifier
- uon:29258
- Identifier
- ISBN:9781450340427
- Language
- eng
- Reviewed
- Hits: 2652
- Visitors: 2894
- Downloads: 1
Thumbnail | File | Description | Size | Format |
---|