Title:  Detection of seal pups and polar bears in multispectral imagery

Presenter(s): Juanita Jimenez,

Date: Friday, February 23 · 1:35 – 2:05 pm EST

Google Meet joining info

Video call link: https://meet.google.com/ejj-kthh-sje 

Or dial: ‪(US) +1 914-893-5689‬ PIN: ‪679 479 638‬#

More phone numbers: https://tel.meet/ejj-kthh-sje?pin=5439818462517

About Speaker:  Juanita Jimenez

Abstract: The Polar Ecosystems Program at the Alaska Fisheries Science Center’s Marine Mammal Laboratory is responsible for estimating the abundance and distribution of ice-associated seals of Alaska. They conducted large-scale fixed-wing surveys of the sea ice in the Bering, Chukchi, and Beaufort seas in a collaborative effort with the US Fish and Wildlife Service to include polar bears in this effort, and international researchers to accurately assess transboundary population estimates. These surveys result in the collection of hundreds of thousands of aerial images of the sea ice from airborne color (RGB), thermal infrared (IR), and ultraviolet (UV) cameras. Machine learning models can be used to identify ringed seals, bearded seals, and polar bears in the imagery. The objective of this NOAA Experiential Research and Training Opportunity, NERTO was to utilize the annotated UV, IR, and color imagery to develop a detection model to find polar bears and white coat seal pups in multispectral imagery. During the summer, we started the development of a processing workflow, first using ENVI and then with MATLAB to process these images. The approach being pursued looks at building a five band multispectral data cube by first registering the images and stacking them into a three-dimensional array. Once the images are aligned, animal signatures can be used to detect the presence of mammals in the scene. The goal of the NERTO, was to identify the workflow that NOAA needs to extract spectral signatures using the UV, IR, and color images. The final goal is to build a model to find the animals in the sea ice by processing these cubes. Follow up work will be pursued as part of my master’s project at UTEP.