Hooked on AI: Applying Computer Vision Methods to Remote Camera Images to Create Efficiency Savings in Recreational Fishing Surveys

Updated: 1 day ago
Location: Mount Lawley, WESTERN AUSTRALIA
Deadline: 30 Nov 2022

Project Overview

Remote cameras can provide greater insight into recreational fishery dynamics and assist in the sustainable management of fisheries resources. As such, cameras are increasingly being used to monitor recreational fishing effort in boat-based marine recreational fisheries. Remote cameras can provide ‘complete’ coverage of boating activity (i.e., 24 hrs a day, every day of the year) in contrast to boat ramp surveys. Counts of vessel retrievals from remote cameras can support estimation of recreational fishing effort and provide a proxy for inferring trends in fishing effort between periodic boat ramp surveys. For most remote camera studies, video images are manually checked by readers requiring substantial resourcing. Automated computer vision and artificial intelligence (AI) methods offer potentially cost-effective means to interpret images. Despite the perceived benefits that these innovative approaches would provide, few published studies have applied AI methods to remote camera data in recreational fishing surveys.

This PhD project will investigate and develop efficient computer vision algorithms to analyse remote camera data to classify video images by vessel type and provide counts of recreational vessels.


At a glance

The Department of Primary Industries and Regional Development has video images from an extensive network of remote cameras across the State. These cameras monitor vessel traffic at boat ramps, choke points, estuary channels and groynes with the potential to develop a census of recreational vessel activity across time. Data are currently processed by video image readers who identify and count vessels manually. The cost and time requirements to process data in this manner limit the ability to read all data, leading to data gaps. It is therefore desirable to investigate alternative methods for reading and classifying remote camera footage. Video images could be used to train computer vision with the intended outcome being the ability to automate the reading process and provide reliable information on boating activity across the State.


Eligibility guidelines

Applicants should have a high level of achievement, including an Honours degree, Masters by Research or equivalent. The degree could be in a range of relevant fields, such as Data Science, Machine Learning, Artificial Intelligence and Computer Visions.

To be eligible for this scholarship, the applicant much:

  • meet the PhD admission requirements at ECU;
  • be able to demonstrate strong background and competencies in data analysis, machine learning and/or artificial intelligence. Experience in deep learning is ideal;
  • demonstrate excellent programming in Python;

How to Apply

Interested applicants are encouraged to submit an expression of interest to Dr Johnny Lo.

Please ensure the below documents are attached to the email:

  • a cover letter that includes a brief statement of your suitability with respect to the above key requirements;
  • a curriculum vitae (CV), including your academic transcripts and a list of your published research (if any);
  • evidence of English-language proficiency (only relevant for applicants whose previous studies were not in English);
  • any supporting documents of prior research experience in relevant areas.

Expressions of interest will only be accepted via email.


Stipend

The successful applicant will receive a scholarship of $30,000.00 per year for 3 years pending compliance with milestone and candidacy requirements.


For more information

Contact j.lo@ecu.edu.au


View or Apply

Similar Positions