Fish Degradation Early Alert System

This application is designed to address the critical challenge of fish spoilage detection in cold storage environments. Using sensor data collected at regular intervals, the system applies time-series analytics including rolling statistics, change point detection to identify early signs of degradation.

Users can monitor multiple sensors (e.g., ammonia, hydrogen sulfide) across different devices in real time. A smart fusion algorithm combines signals from multiple methods and sensors to generate probabilistic alerts. The intuitive dashboard provides clear visualizations of trend deviations and detected change points to enable timely interventions.

The solution improves food safety, reduces economic losses, and is especially useful in aquaculture supply chains, fish markets, and cold logistics environments.

Key Differentiators

• Real-time Multi-Sensor Analysis: Continuously monitors gas emissions (like ammonia, hydrogen sulfide) every 5 minutes, enabling timely detection of spoilage trends before they become critical.
• Intelligent Probability Fusion Engine: Combines outputs from multiple anomaly detection methods (z-score, rate of change, rolling statistics) across multiple sensors into a single soft-probability alert, minimizing noise and maximizing reliability.
• Early Warning, Not Just Detection: Goes beyond detecting outliers by identifying gradual degradation patterns, offering hours of lead time for corrective action. 
• Dynamic Visualization Dashboard: Intuitive, real-time plots show sensor trends, method-specific scores, and combined probability alerts — enabling quick diagnostics and decision-making.
• Configurable & Extensible: Sensor thresholds, detection methods, fusion rules, and visualization options can all be tailored per use case, device, or storage environment.
• Optimized for Cold Chain Logistics: Designed for real-world variability in cold storage — handles missing data, irregular sensor behavior, and non-stationary trends robustly.
• No Additional Hardware Required: Works with existing sensor infrastructure — no need for high-end cameras or lab testing — making it highly cost-effective and scalable.