Related Experiment Video
Updated: Jan 13, 2026
14:03
Using an Automated 3D-tracking System to Record Individual and Shoals of Adult Zebrafish
Published on: December 5, 2013
11.4K
A Multi-Fish Tracking and Behavior Modeling Framework for High-Density Cage Aquaculture.
Xinyao Xiao1,2,3, Tao Liu1,4, Shuangyan He1,2,3
1State Key Laboratory of Ocean Sensing & Ocean College, Zhejiang University, Zhoushan 316021, China.
Sensors (Basel, Switzerland)
|January 10, 2026
Summary
SOD-SORT improves multi-fish tracking in deep-sea cages by integrating advanced motion models with appearance features. This novel approach reduces identity switches and enables reliable behavior analysis for aquaculture monitoring.
Area of Science:
- Computer Vision
- Animal Behavior Analysis
- Aquaculture Technology
Background:
- Multi-fish tracking in deep-sea environments is challenging due to homogeneous fish appearance and low image quality.
- Standard linear motion models are inadequate for complex, nonlinear fish swimming patterns, causing tracking errors.
Purpose of the Study:
- To enhance multi-fish tracking and behavior analysis in aquaculture settings.
- To address limitations in appearance-based association and motion modeling for fish tracking.
Main Methods:
- Proposed SOD-SORT, integrating a Constant Turn-Rate and Velocity (CTRV) motion model within an Extended Kalman Filter (EKF) into the DeepOCSORT tracker.
- Employed Bayesian optimization for EKF parameters (Q, R) and ReID weighting.
- Introduced statistical quantization for trajectory analysis and behavior classification.
Main Results:
- SOD-SORT achieved an IDF1 score of 0.829 and reduced identity switches by 13% on the DeepBlueI dataset.
- Optimized integration resolved performance degradation from naive CTRV-EKF implementation.
- Successfully classified normal and abnormal swimming behaviors using quantized trajectories on multiple datasets.
Conclusions:
- Principled integration of advanced motion models and appearance cues enhances fish tracking reliability.
- High-quality, continuous trajectories are crucial for accurate behavior modeling in aquaculture.
- The proposed method offers a robust solution for monitoring fish welfare and behavior in marine environments.
