Deep learning for underwater scene classification

Ammar Mahmood

Research output: ThesisDoctoral Thesis

638 Downloads (Pure)


This thesis poses the underwater image analysis as challenging computer vision problems and aims to address those using deep learning techniques. It proposes state-of-the-art deep learning methods to automatically classify coral reefs and kelp forests. It also extends these methods to automatically depict the trends of changing coral population and coverage of kelp forests. Lastly, the thesis proposes to generate synthetic parts data for automatic detection of lobsters in the complex underwater background, a challenging task in the marine research community.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • The University of Western Australia
  • Hovey, Renae, Supervisor
  • Sohel, Ferdous, Supervisor
  • Bennamoun, Mohammed, Supervisor
  • An, Senjian, Supervisor
Award date25 Mar 2019
Publication statusUnpublished - 2019


Dive into the research topics of 'Deep learning for underwater scene classification'. Together they form a unique fingerprint.

Cite this