Data Science MS Thesis Defense - "FishNet Monitor: An Edge-Optimized Framework for Automated Fishing Activity Detection in Maritime Surveillance", by Yingzhe Qin
"FishNet Monitor: An Edge-Optimized Framework for Automated Fishing Activity Detection in Maritime Surveillance"
Date: Monday, April 28, 2025
Time: 10am-12pm
Via Zoom:
Join Zoom Meeting
https://umassd.zoom.us/j/99701504547?pwd=2cIaHmbfAJZCBDADbDuX4qoIQjddkY.1
Meeting ID: 997 0150 4547
Passcode: 975098
Committee Chair: Ming Shao, Miner School of Computer & Information Sciences, UMass Lowell
Committee Members:
Bo Dong, Mathematics Department, UMass Dartmouth
Donghui Yan, Mathematics Department, UMass Dartmouth
Abstract
This thesis presents FishNet Monitor, a resource-efficient framework for detecting fishing net deployment and retrieval actions in long, continuous video streams from commercial fishing vessels. Maritime monitoring systems face significant challenges in edge deployment scenarios where computational resources are limited, yet accurate fishing action detection remains critical for regulatory compliance and sustainable resource management. We address these challenges through a systematic approach that combines feature engineering, algorithmic innovation, and sampling optimization tailored specifically for maritime environments. First, we demonstrate that Average Detections Per Frame (ADPF) provides a more effective feature than bounding box aspect ratios for distinguishing between consecutive fishing actions in challenging sea conditions.Second, we develop an adaptive peak detection algorithm that robustly identifies fishing events despite variable time intervals between them. Third, we implement and evaluate two sampling strategies—fixed and dynamic—across 152 experimental configurations to optimize the inference cost versus accuracy tradeoff for extended maritime operations. Our results demonstrate that afixed sampling approach with frame skip=4 achieves 100% detection accuracy while reducing inference costs to 25% of full-frame processing, enabling continuous monitoring throughout extended fishing expeditions. Further analysis reveals that while both sampling strategies deliver comparable inference cost savings, the fixed approach provides significant runtime advantages for edge deployment on fishing vessels. Performance projections indicate the framework canprocess maritime surveillance footage in real-time on the Hailo 8 edge computing platform,offering a practical solution for automated fishing action detection that supports compliance monitoring, sustainable fishing practices, and fleet operation optimization.
For additional information, please contact Ming Shao at Ming.Shao@uml.edu
Via Zoom: Join Zoom Meeting https://umassd.zoom.us/j/99701504547?pwd=2cIaHmbfAJZCBDADbDuX4qoIQjddkY.1 Meeting ID: 997 0150 4547 Passcode: 975098