SARTAB: A System for Real-Time Behavior Detection in Cichlid Fish
Author Information
Author(s): Lancaster Tucker J., Leatherbury Kathryn N., Shilova Kseniia, Streelman Jeffrey T., McGrath Patrick T.
Primary Institution: Georgia Institute of Technology
Hypothesis
Can a scalable system for automated real-time behavior detection improve the analysis of courtship behavior in cichlid fish?
Conclusion
The SARTAB system successfully detects courtship behaviors in cichlid fish with high accuracy and efficiency.
Supporting Evidence
- SARTAB achieved high ROI and animal detection accuracies with mAP@[.5:.95] of 0.969 and 0.718, respectively.
- The system demonstrated 100% classification accuracy on a set of 32 manually selected behavioral clips.
- Real-time analysis was facilitated using low-cost, edge-deployed hardware.
Takeaway
Researchers created a smart system that watches fish and tells them when they are courting, helping them study fish behavior better.
Methodology
The study used a modular system with a Raspberry Pi and Edge TPU to monitor fish behavior in real-time, analyzing video data to detect courtship.
Potential Biases
Potential misidentification of dither fish as P. demasoni could affect detection accuracy.
Limitations
The system's performance may be limited by the computational power of Edge TPUs compared to GPUs.
Participant Demographics
The study involved sexually mature Pseudotropheus demasoni cichlid fish.
Statistical Information
P-Value
0.001
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website