University of Surrey

Test tubes in the lab Research in the ATI Dance Research

Binary Images of Real-World Nuclear Waste Simulants for Visual Object Recognition

Shaukat, A and Gao, Y Binary Images of Real-World Nuclear Waste Simulants for Visual Object Recognition [Dataset]

Text (licence)
Available under License : See the attached licence file.

Download (33kB) | Preview
Item Type: Dataset
Divisions : Surrey research (other units)
Principal Investigator :
Principal InvestigatorEmail
Description : The dataset was generated using real-world nuclear waste simulants provided by the National Nuclear Laboratory and Sellafield Ltd. UK. Images recorded using the perspective projections of each simulant were distinctly placed on a sorting table for multiple values of ψ and φ. In order to maximise the disparity in object orientation within the training/test sample space, a secondary subsampled dataset is generated from the original set of projections by rotating each image over the angular domain of θ. Each rotated image is further scaled by reducing or increasing by {1.1, ..., 1.5} × original image, which creates a stack of successively smaller and larger images to simulate size variations. Together, they constitute a superset of 86400 subsampled images of training and test data with ground truth annotations from which, in principle, maximally discriminative mappings of 2-D shape projections to distinct class labels can be derived. This dataset further provides the ability to train and test the system in terms of varying object positions and orientations on the sorting table. Collection of data comprising objects at different positions, orientations and locations on the sorting table helps in training and testing the algorithm with anomalies that may arise due to changes in the perspective camera projections of the objects and the resulting effects in their binary silhouettes [Refer to related paper].
Publication Year of Data : 2015
Funder : Engineering and Physical Sciences Research Council (EPSRC)
DOI : 10.15126/surreydata.00808553
Grant Title : Reconfigurable Autonomy
Access Statement : Acknowledgement required for usage of data set: Surrey Space Centre (STAR Lab), University of Surrey, Guildford, GU2 7XH, U.K.
Data Access Contact :
Data Access ContactEmail
External Data Location : Internal Location
Keywords : Robot vision systems, Object recognition, Automatic optical inspection, Nuclear decommissioning, Industrial automation, Sort-and-segregate
Depositing User : Symplectic Elements
Date Deposited : 09 Sep 2015 08:12
Last Modified : 23 Jan 2020 09:54

Actions (login required)

View Item View Item


Downloads per month over past year

Information about this web site

© The University of Surrey, Guildford, Surrey, GU2 7XH, United Kingdom.
+44 (0)1483 300800