Active Learning Training System - AquaWatch Environmental Intelligence
AquaWatch Environmental Intelligence (AEI) uses an active learning training system for quickly training custom image classification models. This system gives us complete control on training models with classes agreed with the customer, and fine-tuning models as additional data becomes available.
Our modular approach allows us to provide layered solutions alongside the specific CSO incidence monitoring. With the potential to include water level, flow and other desired monitoring algorithms as well as the physical water parameters gathered in real time by the AquaWatch Waka.
The Active Learning Journey:
Initial Data with Impact: The journey begins with the AW-108’s initial data capture, recording the baseline of environmental conditions. These details can be shared immediately, meaning that monitoring and learning can go hand in hand, adding value from the initial deployment.
The Feedback Loop: Through iterative cycles, AEI selects the most informative images to be annotated, enhancing its detection algorithms.
Strategic Sampling: Utilising strategies like uncertainty and diversity sampling, the AW-108 zeroes in on the data that will most improve its performance.
Real-World Applications: From monitoring trash racks to managing water levels, the AW-108 actively adapts, applying its growing intelligence to a range of environmental challenges.
AquaWatch Environmental Intelligence (AEI) Process
-
Defining the Problem:
We start by identifying specific environmental challenges, such as Combined Sewer Overflows (CSOs), trash rack fouling, or discharge points, that require vigilant monitoring and swift response.
-
Data Gathering:
Utilising the AW-108’s 4K video capture, we collect high-fidelity visual data from various water systems, ensuring a comprehensive dataset that captures the nuances of environmental changes. Depending on the solution required, this can start from 100-1000 images but monitoring is possible during the data gathering process, adding immediate value from the hardware investment.
-
Model Training:
With expert annotation of this visual data, we train robust machine learning models that can accurately detect early signs of environmental issues like CSOs.
-
Outcome Refinement:
Our models are refined through iterative training to pinpoint desired outcomes, such as reducing pollution levels and mitigating health risks.
-
API and Insight Delivery:
Insights gleaned from the data can be delivered via a user-friendly API, equipping stakeholders with real-time data to make informed decisions.
-
Ongoing Calibration and Support: The AW-108 system is designed for continuous improvement. With each event, our models learn and adapt, reducing the need for human intervention and enhancing precision over time.
Real World Examples
In each of these scenarios, AEI enables a more adaptive, responsive, and efficient environmental monitoring system, leading to better resource management and environmental protection.
Trash Rack Monitoring
In the context of trash rack monitoring, active learning is used to refine the detection of debris accumulation in real time.
The AW-108 collects visual data on the state of trash racks, and over time, the system actively identifies and learns the patterns of debris build-up. As a result, maintenance teams can be alerted more efficiently to prevent blockages, reduce the risk of flooding, and maintain water flow.
GPT (Gross Pollutant Traps) Monitoring
For gross pollutant traps, active learning helps in identifying when traps are nearing capacity or malfunctioning.
The AW-108 AI camera can monitor the trap, and through active learning, it becomes adept at recognizing signs that a clean-up or repair is necessary. This ensures that pollutants are consistently captured and removed, preventing them from entering water bodies.
CSO (Combined Sewer Overflow) Monitoring
The AW-108 AI camera is an ideal solution for monitoring CSOs. Through active learning, the camera can detect and predict overflow events by learning from past incidents.
It recognises patterns such as rapid water level rises, unusual water coloration, or increased turbidity. This proactive approach allows for timely interventions to mitigate the environmental impact of overflows.
Water Flow and Level Monitoring
Active learning plays a critical role in managing water resources by monitoring flow and water levels using cameras like the AW-108. It can predict potential flooding or drought conditions by learning normal and atypical patterns.
Over time, the system becomes increasingly accurate in alerting authorities to unusual changes, which is crucial for water management and disaster prevention.