Document Type
Dissertation
Disciplines
1.2 COMPUTER AND INFORMATION SCIENCE, Computer Sciences
Abstract
Computer vision is slowly becoming more and more prevalent in daily life. Tesla has recently announced that it plans to scale up the manufacturing of their Robotaxis by 2024, with this increase in self-driving vehicles being just one example, the importance of computer vision is growing year by year. Vision can be easy to take for granted, as most humans grow up using vision as their primary way of absorbing environmental information. The way humans process and classify visual information differs significantly from how current computer vision systems process and organise visual information. The human brain can use its past knowledge and experience to draw conclusions that would be impossible for a limited artificial system. The human brain has a lifetime of visually informed learning to fall back on when classifying an object, the function of an object, and the relationships between particular objects. Humans can recall this information when needed once initially learned. Current convolutional neural networks (CNN) focus on processing the image at a pixel level, so they do not have this historical and relational information about these objects to fall back on, at least not explicitly. A CNN may connect certain features or objects during the classification process but this is a black box, and the end user has now way of determining this links.
Recommended Citation
O'Neill, A. (2022). KG-CNN: Augmenting Convolutional Neural Networks with Knowledge Graphs for Multi-class image classification. [Technological University Dublin].
Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 International License.
Publication Details
A dissertation submitted in partial fulfilment of the requirements of Technological University Dublin for the degree of M.Sc. in Computing (Data Science) June 2022.