Pixelated Semantic Colorization
- Submitting institution
-
Aberystwyth University / Prifysgol Aberystwyth
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 38748771
- Type
- D - Journal article
- DOI
-
10.1007/s11263-019-01271-4
- Title of journal
- International Journal of Computer Vision
- Article number
- -
- First page
- 818
- Volume
- 128
- Issue
- -
- ISSN
- 0920-5691
- Open access status
- Compliant
- Month of publication
- December
- Year of publication
- 2019
- URL
-
-
- 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
-
3
- Research group(s)
-
-
- Citation count
- 3
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Inspired by the observation of human beings perceiving and distinguishing colours based on semantic categories of objects, this paper designs novel deep Neural Networks that employ pixelated object semantics to guide image colourisation. For the first time, colourisation is viewed as a sequential pixel-wise colour distribution generation task, rather than a pixel-wise classification task, allowing a network to be trained via co-optimising colourisation and semantic segmentation. After presentation at BMVC’18 (acceptance rate <5%), this substantially extended paper was invited and accepted by this No. 1 Computer Vision journal. Conference paper is not returned.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -