Continuous perception for deformable objects understanding
- Submitting institution
-
University of Glasgow
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-01393
- Type
- D - Journal article
- DOI
-
10.1016/j.robot.2019.05.010
- Title of journal
- Robotics and Autonomous Systems
- Article number
- -
- First page
- 220
- Volume
- 118
- Issue
- -
- ISSN
- 0921-8890
- Open access status
- Compliant
- Month of publication
- August
- Year of publication
- 2019
- URL
-
http://eprints.gla.ac.uk/187631/
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 2
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- ORIGINALITY: First demonstration of continuous machine perception of real-time RGBD (colour+depth) dynamic appearance of deformable objects (clothing) for classification purposes, significantly advancing machine vision image understanding over prior single-frame methods. SIGNIFICANCE: Allows the class, type or properties of deformable material to be inferred by observation of intentional physical interaction, i.e. robotic manipulation. Direct application to robotic clothing manipulation tasks, advanced manufacturing processes and visual understanding of dynamic environments in autonomous systems. RIGOUR: Validation on first dynamic annotated clothing database of 150 garment RGBD videos (15 clothing items, 5 categories) that improved the state-of-the-art in clothing recognition by >27%.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -