Alex is a fourth-year student in the combined B.S./M.S. program of the Computer Science & Engineering Department. His research interests lie in the applications of machine learning and computer vision and their intersections within other fields. He is currently working on extending a vision and detection pipeline leveraging drone footage capture as input. Outside of school, he enjoys boxing, skiing, and soccer.
Publications
Detecting and Counting Salmon Redds via Aerial Imagery: An Automated Computer Vision Approach Towards Ecosystem Preservation
Zhang, A.,
Kirby, G.,
Goble, I.,
Lenon, D.,
Koger, B.,
Berdahl, A.,
and Leung, K.
In IEEE Conf. on Computer Vision and Pattern Recognition: CV4Animals: Computer Vision for Animal Behavior Tracking and Modeling,
2025
Surveying salmon spawning nests, or redds, is a critical task for estimating population size, guiding conservation efforts, and understanding habitat use within river ecosystems. Traditional methods rely on manual detection and counting, which are time-consuming, costly, and prone to inconsistencies due to human error and environmental variability. We propose an automated and accessible pipeline for salmon redd detection and counting using oriented object detection on aerial RGB drone footage. Our approach addresses key challenges in redd surveying, including complex river topologies, varied redd orientations, and dense redd clustering. We apply targeted data augmentations for lighting, shading, and water turbidity to improve robustness and generalizability, and incorporate motion-aware tracking with spatial-temporal clustering to reduce duplicate and fragmented detections. Our method achieves 79.27% mAP in rotated detection and significantly reduces manual workload, offering an effective tool for large-scale ecological monitoring and salmon population assessment.
@inproceedings{ZhangKirbyEtAl2025,author={Zhang, A. and Kirby, G. and Goble, I. and Lenon, D. and Koger, B. and Berdahl, A. and Leung, K.},booktitle={{IEEE Conf.\ on Computer Vision and Pattern Recognition}: {CV4Animals: Computer Vision for Animal Behavior Tracking and Modeling}},title={{Detecting and Counting Salmon Redds via Aerial Imagery: An Automated Computer Vision Approach Towards Ecosystem Preservation}},year={2025},img={ZhangKirbyEtAl2025.png},owner={karenl7}}