A unity-based farmed fish target tracking model training data generation method
By simulating the real movement behavior of fish in a virtual environment, a highly diverse and accurate training dataset is generated, solving the problems of high data acquisition costs and large annotation workload in intelligent monitoring of aquaculture, and realizing the development of a low-cost and high-efficiency multi-target fish tracking algorithm.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA AGRI UNIV
- Filing Date
- 2026-04-16
- Publication Date
- 2026-07-24
AI Technical Summary
In intelligent monitoring of aquaculture, existing technologies suffer from high data acquisition costs, large annotation workload, insufficient data volume, and poor scene controllability, which restricts the development and optimization of multi-target tracking algorithms for fish.
This system simulates the real movement behavior of fish in a virtual environment, reproducing the swimming and interaction scenes of fish in the water through 3D simulation technology. It uses an automatic annotation mechanism to generate annotated datasets, including fish model simulation modeling, simulated water environment construction, fish behavior state modeling, FishBrain behavior control module, FishMove motion execution module, FishSpawner batch generation and ID binding module, and MOTRecorder data recording and export module, to achieve automatic recording and output of image sequences and trajectory label data.
It reduced the cost of manual annotation, generated highly diverse training data, increased the amount and accuracy of data, reduced equipment and site construction costs, and improved the development efficiency of intelligent aquaculture systems.
Smart Images

Figure CN122049571B_ABST