Exploring synthetic Image generation for training computer vision models under data scarcity
Moreu, EnricORCID: 0000-0002-0555-3013
(2024)
Exploring synthetic Image generation for training computer vision models under data scarcity.
PhD thesis, Dublin City University.
This thesis presents research conducted in the area of synthetic data generation
for computer vision tasks. The research aims to address the challenge of datahungry deep learning models by generating synthetic images that can effectively train
computer vision models to solve tasks such as object counting, polyp segmentation,
and pattern classification. The work carried out explores the use of various techniques
to ensure effective use of synthetic data, including domain randomisation and domain
adaptation in both self- and semi-supervised setups. Through the application of
these techniques, the research aims to develop a robust and effective approach for
generating synthetic data that can improve the performance of computer vision
models with a reduced amount of human annotations.