11/29/2023
Join IIDS for the Fall Seminar Series
Thursday, November 30th, 12:30 pm
Buchanan Engineering Lab 205
Boyu Zhang
Asst Professor - Dept of Computer Science
U of I College of Engineering
"AI-Driven Automatic Wildlife Health Monitoring"
The widespread use of camera traps has led to an abundance of wildlife imagery, presenting a challenge in effectively utilizing this large data volume. While progress has been made in animal detection and classification from these images, a gap remains in detailed, fine-grained analysis. We proposed to enrich the detailed information of the camera trap images by estimating animal postures to identify key body parts accurately and, thus, perform fine-grained analysis by introducing insights from wildlife professionals. This expert input is crucial, especially when working with smaller scale training datasets, allowing us to achieve more precise analysis.
Via expert-guided feature fusion, the proposed method significantly improves the accuracy of differentiating between mule deer and white-tailed deer species. Furthermore, we augmented with expert knowledge to effectively estimate the fat thickness in wild mule deer. This approach has been shown to surpass the accuracy of traditional, experience-based human visual assessments. Such advancements are key in monitoring the health of wild mule deer populations, contributing to the development of effective conservation policies, and aiding in the early detection of diseases. The proposed method's adaptability to various wildlife species highlights its potential to enhance wildlife conservation efforts substantially.
See full seminar schedule at https://www.iids.uidaho.edu/events.php
