Fishing Forecast Guardian: A Machine Learning-Powered Platform for Illegal Fishing Detection

Abstract

Illegal, Unreported, and Unregulated (IUU) fishing threatens marine ecosystems, food security, and lawful fisheries. This paper introduces ”Fishing Forecast Guardian,” a real-time web-based system that detects illegal fishing activities using AIS data, satellite imagery, and machine learning models. By identifying loitering vessel behavior, the system flags suspicious activity using Random Forest, One-Class SVM, and CNNs. The platform includes an interactive frontend built with React.js and Leaflet.js, and a Python-based ML backend with Flask. This integrated, scalable system aids governmental and environmental stakeholders in monitoring IUU activities efficiently.

Introduction

Problem and Motivation
Illegal, unreported, and unregulated (IUU) fishing causes $10–23 billion in annual losses. Traditional surveillance methods are ineffective due to the ocean's size and tactics like AIS (Automatic Identification System) disabling. There's a need for automated systems that process vast maritime datasets in real time to detect and predict illegal fishing behavior.

System Overview
The Fishing Forecast Guardian is a modular, AI-driven platform that integrates real-time AIS and satellite data with machine learning (ML) to detect loitering and IUU fishing patterns. Its architecture comprises:

1. Data Ingestion & Preprocessing

2. Model Training & Inference

3. Visualization Dashboard

4. Continuous Learning Engine

Key Features & Methodology

• Data Processing: AIS, GPS, and behavioral data are cleaned, normalized, and enriched with features like speed, loitering, and proximity to shore. Noise filtering and data augmentation (e.g., synthetic loitering patterns) improve quality.

• Machine Learning Models: Models used include Random Forest, SVM, Logistic Regression, Isolation Forest, and CNNs (for satellite imagery). Random Forest and Neural Networks showed the best results.

• Prediction Pipeline: Real-time anomaly detection of loitering, AIS disabling, and prediction of future IUU zones based on spatiotemporal data.

Frontend & Visualization

• Built with React.js and Leaflet.js

• Features:

  • Real-time vessel tracking

  • Color-coded risk indicators

  • Historical route playback

  • Map-based queries

  • Chart-based analytics

Backend and Deployment

• Flask API with RESTful endpoints for model training, prediction, and data uploads

• Uses Docker and Kubernetes for scalability

• Average prediction latency: <300ms

• Cloud-ready: Tested on AWS and Google Cloud

Model Performance

Models were optimized using grid and randomized search with cross-validation. The system supports user-driven retraining for adaptability.

Real-World Validation

• Case study in Southeast Asia flagged 27 IUU cases, 19 of which matched known violations — ~70% real-world precision.

• Loitering analysis showed average dwell time of 3.2 hours in high-risk zones.

Comparative Advantage

Compared to systems like Global Fishing Watch, MarineTraffic, and OceanMind, Fishing Forecast Guardian offers:

• Real-time ML-based detection

• Predictive analytics

• Fully interactive dashboard

• Retrainable models by users

Ethical Considerations

The system avoids biases (e.g., nationality) and focuses on behavior-based evidence. It follows fairness, transparency, and accountability principles.

Conclusion

Fishing Forecast Guardian demonstrates how machine learn- ing and web technology can combat IUU fishing at scale. Future improvements include adding environmental data (cur- rents, temperature), expanding geographic scope, and enabling alerts to maritime authorities. We also aim to integrate external APIs for oceanographic data and leverage federated learning to preserve privacy while retraining models collaboratively. Furthermore, incorporating satellite radar and SAR imagery can help detect vessels that disable AIS entirely. Additionally, our platform could serve as a foundational tool for developing international cooperation portals where nations share insights and collaborate on marine protection strategies. Future research can also explore integrating blockchain to ensure transparency and traceability of vessel activity logs. We envision a globally interconnected surveillance system where each observation strengthens the collective model.

References

[1] Wang, X., Zhang, Y. ”Machine Learning for Real-Time Detection of IUU Fishing”, 2023. [2] Kumar, A., Sharma, S. ”Using AIS Data for IUU Fishing Detection”, 2022. [3] Liu, Y. et al. ”Satellite Imagery for Detecting IUU Fishing”, 2021. [4] Jackson, T., Lee, R. ”Global Fishing Watch: Real-Time Monitoring”, 2023. [5] Siddiqui, T. et al., ”AIS-based Maritime Monitoring with ML Tech- niques,” IEEE Oceans, 2022. [6] Rossi, G. et al., ”Multimodal Data Fusion for Maritime Situational Awareness,” Elsevier, 2021. [7] Petremann, A., ”Combating IUU Fishing with Remote Sensing Tech- nology,” Marine Policy, vol. 134, 2022. [8] Li, J., Tan, W., ”Deep Learning for Vessel Trajectory Prediction using AIS Data,” IEEE Access, vol. 9, pp. 78112–78123, 2021. [9] Schiller, L. et al., ”An Overview of Data-Driven Approaches in Maritime Domain Awareness,” Sensors, vol. 20, no. 24, 2020. [10] De Souza, E. et al., ”Detecting Fishing Activity from Vessel Tracks,” Proceedings of the National Academy of Sciences, vol. 113, no. 12, pp. 3124–3129, 2016. [11] Kroodsma, D. A. et al., ”Tracking the Global Footprint of Fisheries,” Science, vol. 359, no. 6378, pp. 904–908, 2018. [12] McCauley, D. J. et al., ”Ending Hide and Seek at Sea,” Science, vol. 351, no. 6278, pp. 1148–1150, 2016. [13] Miller, N. A. et al., ”Identifying Global Patterns of Transshipment Behavior,” Frontiers in Marine Science, vol. 5, 2018. [14] Mazurek, R. et al., ”The Role of Big Data in Ocean Governance,” Marine Policy, vol. 121, 2020. [15] Matwin, S., et al. ”Data Mining Methods for Maritime Surveillance,” Proceedings of the ECML PKDD 2012. [16] Yu, H., Wang, J., ”Anomaly Detection in Maritime Trajectories Using Autoencoders,” Journal of Navigation, vol. 74, no. 5, pp. 1001–1017, 2021. [17] Zhang, Q., Liu, F., ”Ship Detection in Satellite Imagery with Deep Learning Techniques,” Remote Sensing, vol. 12, no. 9, 2020. [18] Kim, D. et al., ”Real-time Vessel Monitoring using Edge AI for Smart Ports,” IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 2, pp. 1180–1192, 2022. [19] Elahi, M. et al., ”AIS Data Mining for Maritime Surveillance and Intelligence,” Journal of Marine Science and Engineering, vol. 9, no. 4, 2021.

Copyright

Copyright © 2025 Pankhuri Sharma, Mohit Panjikar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.