←back to projects
Dec 2024/AI/ML/competition

AquaGuard

AquaGuard focused on non-intrusive pool-safety monitoring by using YOLO-based detection on camera streams to surface high-risk events earlier.

Domain
AI/ML
Role
Machine Learning Engineer
Output
ML Pipeline
Category
Computer Vision Safety System
Project Framing

A source-backed case study built for recruiter review

This reading path makes the problem choice, evidence quality, user framing, execution decisions, and proof trail visible without overstating what the sources support.

Project Type
competition

Gold-medal iCE-CInno 2024 project delivering a CCTV-compatible computer-vision workflow for real-time drowning-risk detection.

Orientation
Tech

Earned Gold Medal recognition at iCE-CInno 2024 in the university undergraduate category.

Core Stack
Python · YOLO · OpenCV · FastAPI

Inference pipeline based on YOLO models with service interface for near-real-time detection reporting.

Why This Problem Mattered

Problem framing before execution

The case-study layer starts with why this problem was selected and how the context justified investment.

Problem Framing Map

Issue

Pool safety monitoring can fail when high-risk behavior depends on delayed human observation rather than early machine-assisted detection.

Context

AquaGuard is documented as a competition-recognized computer-vision safety concept centered on CCTV-compatible drowning-risk detection.

Why Selected

It is a strong Wave 3 addition because it combines public recognition, safety-critical framing, and real-time-oriented ML system reasoning.

Problem statement

Pool safety monitoring often depends on delayed human observation, increasing critical response risk.

Solution thesis

Developed a computer-vision detection flow to identify high-risk behavior from camera streams and trigger early alerts.

Research and Evidence

What supports the narrative

Evidence is surfaced with its source type and credibility note so the recruiter can quickly see what is directly backed versus intentionally constrained.

Safety-system framing
hybrid

The project explicitly focuses on non-intrusive pool monitoring and high-risk event detection from camera streams.

Credibility: Backed by the current project record and CV-based source references.
Recognition signal
hybrid

The project earned Gold Medal recognition at iCE-CInno 2024.

Credibility: Directly supported by proof and documented project evidence references from documented records.

Credibility Notes

  • ●The case should be framed as safety-oriented ML concept and workflow evidence, not as a deployed life-critical product with field validation.
  • ●No false-positive, recall, or deployment reliability metric is added unless explicitly supported in source.
Who The User Was

User framing stays explicit

When formal research artefacts are not available, the page still explains who the work served and why that user framing is justified by the existing sources.

Primary user
Pool operators or safety teams who need earlier visibility into potential drowning-risk events.

The problem statement and solution framing explicitly revolve around earlier alerting for safety monitoring.

Evaluation stakeholder
Reviewers assessing whether the ML approach is suitable for CCTV-compatible safety use cases.

The strongest public signal comes from competition recognition and real-time-oriented system framing.

Decision Flow

How design thinking translated into decisions

The goal is to show the trace from research and insight to concrete product or system decisions, then to the outcomes those decisions supported.

Design Thinking Flow

Each step keeps the movement from evidence to action explicit before the rationale expands it.

  1. Step 1
    Safety-risk framing

    Defined the project around earlier risk detection rather than generic object detection performance.

    Signal: Response urgency became the central product need.
  2. Step 2
    Camera-compatible approach

    Chose a non-intrusive monitoring direction using visual streams instead of requiring wearable or manual input signals.

    Signal: The system remained aligned with practical CCTV deployment contexts.
  3. Step 3
    Recognition-backed packaging

    Turned the concept into a reviewable competition artefact with clear safety-system narrative.

    Signal: Award context strengthens credibility without overstating real-world rollout.

Decision Rationale

Each decision keeps the path from insight to execution visible before ending on the outcome signal.

YOLO for real-time constraints
Insight

Safety monitoring loses value when inference is too slow for practical response windows.

Decision

Used a YOLO-based detection approach to match near-real-time monitoring needs.

Outcome

The project narrative stays aligned with speed-sensitive safety use cases.

Non-intrusive monitoring
Insight

Pool environments benefit from monitoring approaches that do not depend on active user participation.

Decision

Framed the solution around CCTV-compatible computer-vision detection.

Outcome

The product direction becomes easier to understand as a practical safety workflow.

Solution and System Execution

Execution choices and delivery details

This section preserves the technical and operational substance: architecture, responsibilities, trade-offs, and implementation quality signals.

System Design

Inference pipeline based on YOLO models with service interface for near-real-time detection reporting.

Source-backed Impact

Earned Gold Medal recognition at iCE-CInno 2024 in the university undergraduate category.

Responsibilities

  • ●Built model workflow and evaluation process
  • ●Collaborated on deployment-ready inference path

Stack Decisions

  • ●Used YOLO-based approach for real-time-oriented detection constraints

Trade-offs

  • ●Accepted model complexity to improve detection sensitivity

Challenges

  • ●Balancing false positives with safety-critical recall
Outcomes and Proof

What was delivered and what can be verified

Outcome claims remain conservative and source-backed, while proof records and recruiter-safe links surface the strongest verification trail available.

Validation Signals

  • ●Gold Medal iCE-CInno 2024 is recorded as project proof.
  • ●The project record documents a real-time CCTV-compatible drowning-risk detection workflow.

Source-backed Outcomes

  • ●Gold Medal iCE-CInno 2024
  • ●Real-time CCTV-compatible drowning-risk detection workflow
Retrospective and Limits

What the project proves, and what it does not

Strong case studies show both what was learned and where the current evidence stops.

Retrospective

A production path should include an incident-labeling feedback loop for continuous model improvement.

Evidence Limits

  • ●Current sources do not provide field deployment evidence, calibrated safety metrics, or operational incident outcomes.
  • ●The project should remain framed as competition-recognized safety-system reasoning and ML workflow evidence.

Lessons

  • ●Safety products require rigorous scenario-based validation