Authors:
Sheng Yan
and
Jo Arve Alfredsen
Affiliation:
Norwegian University of Science and Technology, Norway
Keyword(s):
Infrared Depth Camera, Behavior Analysis, European Lobster, Animal Tracking.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Image and Video Analysis
;
Motion, Tracking and Stereo Vision
;
Segmentation and Grouping
;
Video Surveillance and Event Detection
Abstract:
European lobster is a highly treasured seafood, but aquaculture production based on traditional communal
rearing practices has proved challenging for this species due to its inherent agonistic behavior. This paper
presents a novel computer vision system that is designed for analysis of lobster behavior and can serve as a
tool to assist selection of breeding stock in a prospective selective breeding program for European lobster.
The automated tracking system provides large quantities of behavioral data for boldness and aggressiveness
analysis, and the infrared light source causes less disturbance to the nocturnal animal under observance. In
addition, because the object is recognized based on depth information instead of color or grayscale pattern
recognition, there are no restrictions on the selection of color or material for the substrate in the experimental
setup. This paper also contributes towards diminishing tracking error caused by water surface reflection and
robust body orient
ation estimation in case of inaccurate body segmentation. We tested ten European lobsters
sized between 25-30 cm to demonstrate the performance and effectiveness of our proposed algorithm.
(More)