Invited Speakers

Dr. Kim Juniper

Ocean Networks Canada & School of Earth and Ocean Sciences

University of Victoria, Victoria, BC, Canada

Extracting ecological information from deep-sea imagery

Abstract: 

The development of tools for the acquisition of deep-sea imagery has surpassed the development of tools for extracting biological information from growing video and photo archives. Yet these imagery archives contain a wealth of information on the spatial and temporal distribution of deep-sea species, and their natural dynamics. Ocean Networks Canada has been operating cabled deep-sea observatories in the northeast Pacific since 2006. Traditional manual extraction of biological information from ONC’s archived imagery has revealed the potential power of time series biological observations in understanding the response of deep-sea species to short- and long-term environmental variability. However, as imagery accumulates in the archive, it is becoming impossible for researchers to exploit the full potential of the thousands of hours of video from multiple deep-sea cameras connected to the network. Ocean Networks Canada has begun developing other solutions from extracting biological information from underwater imagery. This include working with computer scientists to develop machine vision tools that recognize and enumerate species, and the development of a crowd-sourcing tool that allows members of the public to annotate segments of archived video imagery, with information about species present and habitat properties.  This presentation will provide an overview of recent research results that have used Ocean Networks Canada imagery combined with oceanographic sensor data to study seafloor ecosystem dynamics. The presentation will conclude with a review of a recent study that compared crowd sourcing results with expert annotation and computer algorithms to quantify the abundance of sablefish in a submarine canyon habitat.

Bio:

Dr. Kim Juniper is Chief Scientist with Ocean Networks Canada, a University of Victoria-based organization that operates cabled ocean observatories in the Pacific and Arctic Oceans. He is also Professor in UVic’s School of Earth and Ocean Sciences, and holder of the British Columbia Leadership Chair in Ocean Ecosystems and Global Change. He has authored more than 120 peer-reviewed publications on the microbiology, biogeochemistry and ecology of deep-sea hydrothermal vents and other marine habitats. He pioneered the early use of deep-sea video imagery for the study of the natural dynamics of cold seep and hot vent ecosystems in deep ocean trenches and on underwater volcanic ridges.  He has provided occasional scientific expertise to the International Seabed Authority since 2000, in relation to the development of regulations for deep-sea mining, and co-authored the Marine Genetic Resources chapter in the recent UN World Ocean Assessment.  He is a participant scientist and member of the Scientific Advisory Committee for the Canadian Healthy Oceans Network, a government-academia partnership that is researching the ecological importance of biodiversity in Canada’s three oceans.

Prof. Robert B. Fisher FIAPR, FBMVA

School of Informatics

University of Edinburgh, Edinburgh, Scotland, UK

Collecting, Cleaning and Analyzing a Large Fish Dataset

Abstract:

The Fish4Knowledge project captured approximately 1.4 billion images of tropical reef fish from 10 underwater cameras off the coast of Taiwan, over 2010-2013. The project investigated fish detection, tracking and species-level recognition, which ultimately led to the large database. The database was then used for further research, in particular:

1) methods for cleaning the many false positive fish detections, and

2) exploring the relation between fish swimming speed and water temperature (a big noisy data problem).

The talk will cover aspects of both the initial data collection, and also of the post-collection analysis. For fun, we'll also present work on recognizing individual clownfish.

Bio:

Prof. Robert B. Fisher FIAPR, FBMVA received a BS (Mathematics, California Institute of Technology, 1974), MS (Computer Science, Stanford, 1978) and a PhD (Edinburgh, 1987). Since then, Bob has been an academic at Edinburgh University, including a stint as of Dean of Research in the College of Science and Engineering. He has been the Education Committee chair and is currently the Industrial Liaison Committee chair for the Int. Association for Pattern Recognition. His research covers topics in high level computer vision and 3D and 3D video analysis, focussing on reconstructing geometric models from existing examples, which contributed to a spin-off company, Dimensional Imaging. More recently, he has also been researching video sequence understanding, in particular attempting to understand observed animal behaviour. The research has led to 5 authored books and more than 288 peer-reviewed scientific articles or book chapters. He has developed several on-line computer vision resources, with over 1 million hits. Most recently, he is the coordinator of an EU funded project developing a robot hedge-trimmer. He is a Fellow of the Int. Association for Pattern Recognition (2008) and the British Machine Vision Association (2010).