YOLO Salmon Stress Detection
Detecting fish stress through swimming patterns using computer vision
I built a model that analyzes salmon swimming behavior using YOLOv8. The system detects fish in video and visualizes movement patterns to identify potential stress signals. The model is trained on a custom dataset and uses OpenCV for real-time visualization.
See it in action
The model detects fish and tracks movement patterns frame by frame.
📄 Article - Full breakdown
Dataset creation, training process, and model logic explained step by step.
Read article →Tech stack
Status
Completed (Prototype)
Current demo uses substitute footage due to limited access to high-quality salmon videos. I am actively reaching out to obtain real-world data for further validation.
Demo Day Presentation
Presented to 30+ people during Demo Day — a final showcase of projects developed over several months.