Loading...

Media is loading
 

Presentation Type

Presentation

Full Name of Faculty Mentor

Xiangxiong Kong, Physics and Engineering Science

Other Mentors

Erin Burge, Marine Science

Major

Engineering Science

Presentation Abstract

According to the National Ocean Service, only 5% of the oceans have been explored. New behaviors of underwater species call to question how much we even know about what little we have discovered. With so much left to discover, the call for novel methods of marine observation is an urgent research need. The accuracy and availability of the automated observation option through computer vision and image processing have shown great potential to assist marine observation. In this study, we proposed a computer vision-based methodology to program a system to extract, identify, and report fish movements that may not be easily seen by human eyes. Our method has been validated through the MATLAB computer vision toolbox using field images taken from video footage of a live-streaming underwater camera installed beneath Frying Pan Tower in North Carolina. Results indicated our method can successfully identify and track movements of fishes from the video frames.

Location

Room 2 (BRTH 112)

Start Date

12-4-2022 3:00 PM

End Date

12-4-2022 3:20 PM

Disciplines

Engineering Science and Materials

Share

COinS
 
Apr 12th, 3:00 PM Apr 12th, 3:20 PM

Fish identification through video motion tracking from a publicly available live-streaming camera

Room 2 (BRTH 112)

According to the National Ocean Service, only 5% of the oceans have been explored. New behaviors of underwater species call to question how much we even know about what little we have discovered. With so much left to discover, the call for novel methods of marine observation is an urgent research need. The accuracy and availability of the automated observation option through computer vision and image processing have shown great potential to assist marine observation. In this study, we proposed a computer vision-based methodology to program a system to extract, identify, and report fish movements that may not be easily seen by human eyes. Our method has been validated through the MATLAB computer vision toolbox using field images taken from video footage of a live-streaming underwater camera installed beneath Frying Pan Tower in North Carolina. Results indicated our method can successfully identify and track movements of fishes from the video frames.