The aim of this workshop was to inspire interdisciplinary research in the area of automated analysis of deep- sea video data. Data from Ocean Networks Canada were used to illustrate representative examples of research problems in the field. The workshop was targeted at anyone with an interest in cabled observatory data, deep-sea video data, computer vision, video processing and/or marine biology. The workshop ran from 9-12am on October 15 at MTS/ IEEE Oceans ’12 in Hampton Roads, VA.
The first part of this workshop introduced participants to the Ocean Networks Canada data tools, including the data search interface, a data plotting utility, and the deep-sea video archive. Participants were encouraged to bring a laptop to interactively experiment with the tools.
In the second part, we focused in particular on video data. We presented an overview of the main challenges for computer vision and video processing techniques in underwater imaging applications. These challenges include, but are not limited to:
light propagation in underwater media (forward and backward scattering, wavelength-dependent attenuation with distance)
presence of floating particles (marine snow)
analysis of massive amounts of data (HD videos recorded on a 24/7 basis)
A comprehensive survey of the state-of-art techniques in underwater applications of computer vision will also be presented. Among the techniques to be discussed are object recognition, motion analysis, and video summarization. The last part of the workshop will be dedicated to discussion and questions from the audience.