Sorting batteries with deep neural networks - using deep belief networks and convolutionary neural networks for image classification.
Year: 2011
Type: dissertation
Abstract: This master thesis evaluates the possibilities of using dierent versions of deep Articial Neural Networks for identifying batteries at a battery sorting facility.GPU implementations of dierent versions of Deep Belief Networks and Convolutionary Neural Networks are evaluated on data sets made up by images of batteries.Dierent ways to get higher performance from the nets, in terms of percent correct classied, are then explored.The results suggest that the methods could be used to get a fast, and therefore industrial usable system, that can handle the identication of a large number of battery classes, with an error rate within the demands of the industry.1
Author: Rasmus Johansson
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: Trinity: Transcriptome assembly for genetic and functional analysis of cancer, Computational Methods for Gene Discovery, Genome Annotation, and Genome Assembly, Methods for sequencing data analysis and archive-scale data science, Computational Methods for Microbial and Microbiome Sequence Analysis
OpenAlex ID: W36171387