Automatic detection and identification of whales using thermal and visual imaging techniques for cetacean censuses and marine mammal mitigation


Contact
Daniel.Zitterbart [ at ] awi.de

Abstract

AUTOMATIC DETECTION AND IDENTIFICATION OF WHALES USING THERMAL AND VISUAL IMAGING TECHNIQUES FOR CETACEAN CENSUSES AND MARINE MAMMAL MITIGATION Daniel P. Zitterbart, Sebastian Richter, Lars Kindermann, Olaf Boebel For centuries, the spotting of a whale was the apex of days of tedious watch keeping. This skill – lost with the demise of commercial whaling - has recently regained significance, as it forms the basis for scientific cetacean censuses and mitigation efforts of navies and marine geophysical prospection, which both employ loud acoustic sources. We designed, developed, tested and validated an automatic whale detection system (360°) based on a thermal imaging scanner. A coupled visual imaging system allows for identification of automatically detected whales without any marine mammal experts on board. It is capable of detection whales reliably in up to 5 km distance from the ship. Species identification is performed retrospective by a marine mammal expert using high resolution images. We show that this system detects about twice as many whales as visual observers during the same time, and is less effected by environmental conditions like sea-state and wind speed. Moreover is works day and night. It outperforms an alerted observer in terms of number of detections and is a significant step forward in cetacean detection technologies.



Item Type
Conference (Talk)
Authors
Divisions
Primary Division
Programs
Primary Topic
Peer revision
Not peer-reviewed
Publication Status
Published
Event Details
Ocean Science Meeting, Honolulu, Hawaii.
Eprint ID
35712
Cite as
Zitterbart, D. , Richter, S. , Kindermann, L. and Boebel, O. (2014): Automatic detection and identification of whales using thermal and visual imaging techniques for cetacean censuses and marine mammal mitigation , Ocean Science Meeting, Honolulu, Hawaii .


Share

Research Platforms

Campaigns


Actions
Edit Item Edit Item