Deep learning for real-time single-pixel video
- Submitting institution
-
University of Glasgow
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-02580
- Type
- D - Journal article
- DOI
-
10.1038/s41598-018-20521-y
- Title of journal
- Scientific Reports
- Article number
- 2369
- First page
- -
- Volume
- 8
- Issue
- -
- ISSN
- 2045-2322
- Open access status
- Compliant
- Month of publication
- February
- Year of publication
- 2018
- URL
-
http://eprints.gla.ac.uk/155592/
- 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
- 60
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- ORIGINALITY: First work to implement a Deep neural network partially in optics, optimising camera micro-mirror bases to optimally filter single pixel observations and allow real-time solution of the inverse problem, creating useful images faster and more robustly than raster scanning, permitting the first video-rate single-pixel-imaging. RIGOUR: implemented software and hardware, benchmarking performance on large training and test sets. Demonstrated successful 30fps real-time video via a single-pixel on general dynamic scenes different from training data. SIGNIFICANCE: Already high citation impact, as this provides technology for new applications in gas sensing, 3D imaging and metrology which currently have no video sensing arrays.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -