From our research achievements to the technology inside our products — this page shares what we have built and why it works.

Where Acoustics Meets Artificial Intelligence

AIqua's core innovation lies at the intersection of ultrasonic sonar, embedded computing, and deep learning. Our SwimNet sensor array continuously maps the full pool volume with ultrasonic signals — far beyond what any camera can perceive — tracking each swimmer's position, swimming status and drowning risk in real time.

• Acoustic Sonar Array: Multi-element ultrasonic transducers provide full-volume pool coverage to 2.8m depth
• Edge AI Inference: AI model runs on-device — no cloud dependency for core detection, real-time response
• Multi-modal Fusion: Sonar + camera data fusion via the Smart Patrol Box for above-water situational awareness
• Continuous Improvement: Post-deployment on-site fine-tuning and OTA updates keep accuracy at its peak

How SwimNet Sees the Pool

SwimNet replaces the camera with sound. A poolside array of ultrasonic transducers sweeps the entire water volume many times per second, building a live three-dimensional picture of where every swimmer is and how they are moving — from 0.3 m below the surface down to 3 m.

Because the system listens rather than looks, nothing that defeats a camera defeats SwimNet: darkness, surface glare, shadow, reflection and turbid water all leave the acoustic picture intact. The AI model runs on the edge processor at the poolside, so detection never depends on a cloud link.

• Acoustic, not optical — ultrasonic transducers map the full water column, not just the surface
• Behaviour, not appearance — the model flags the change in motion pattern that precedes a drowning
• Nothing recorded — no video, no images, no personal data — only time, zone and alert level
• Works in any condition — unaffected by darkness, glare or water clarity
• On-device inference — core detection runs locally on the AI edge processor, 2–4 sensors each

Smarter Safety from Cameras You Already Have

While SwimNet guards the water, the Smart Patrol Box (SPB-300) watches over the pool deck and surrounding areas. It uses existing IP cameras — ONVIF and RTSP compatible — to detect above-water incidents: slips and falls, fights, intrusions, and unauthorised access.

Together, SwimNet and SPB-300 provide complete 360° coverage of your aquatic facility — from the bottom of the pool to the changing room corridor.

• Uses existing IP cameras — no new hardware investment required
• Edge AI processing — privacy-first face blurring on-device
• Same alert ladder — Pre-Alarm, Warning, Danger unified with SwimNet
• Zoned monitoring — configure detection rules per camera zone
• PDPO compliant — no identifiable footage stored or transmitted

從科研成果到產品核心技術，本頁帶你了解我們的研發歷程。

聲學與人工智能的交匯

AIqua的核心創新在於超聲波聲納、嵌入式運算及深度學習的融合。我們的SwimNet傳感器陣列以超聲波訊號持續掃描整個泳池空間——遠超任何攝影機的感知能力——即時追蹤每位泳客的位置、游泳狀態及溺水風險。

• 聲學聲納陣列：多元件超聲波換能器提供全水池覆蓋，最深達2.8米
• 邊緣AI推理：AI模型在設備本地運行，核心偵測無需依賴雲端，實時響應
• 多模態融合：透過智能巡邏箱整合聲納與攝像數據，提供水面以上的全局監察
• 持續優化：部署後進行現場校調，並透過OTA更新持續提升偵測準確率

SwimNet 如何感知整個泳池

SwimNet 以聲音取代影像。池畔的超聲波換能器陣列每秒多次掃描整個水體，即時建構三維圖像，掌握每位泳客的位置與動作模式——覆蓋水面以下 0.3 米至 3 米。

由於系統靠「聽」而非「看」，凡是令攝影機失效的條件都不影響 SwimNet：黑暗、水面反光、陰影、倒影與混濁水質都不會破壞聲學圖像。AI 模型在池畔的邊緣處理器上運行，偵測完全不依賴雲端連線。

• 聲學而非光學 — 超聲波換能器掃描整個水體，不只是水面
• 判讀行為而非外觀 — 模型偵測的是溺水前的動作模式變化
• 不作任何記錄 — 不錄影、不擷取影像、不儲存個人資料，只保留時間、區域與警報等級
• 全條件運作 — 不受黑暗、反光或水質清濁影響
• 裝置端推理 — 核心偵測在 AI 邊緣處理器本地運行，每個支援 2–4 個感測器

讓現有閉路電視變得更聰明

SwimNet守護水下，智能巡邏箱（SPB-300）則監控泳池池邊及周邊區域。它利用現有IP攝像機——支援ONVIF及RTSP協議——偵測水面以上的事故：滑倒、打鬥、入侵及未經授權進入等。

SwimNet與SPB-300協同運作，為你的水上設施提供從泳池底部至更衣室走廊的全方位覆蓋。

• 利用現有IP攝像機——無需新增硬件投資
• 邊緣AI處理——設備本地人臉模糊，私隱優先
• 統一警報梯級——預警、警示、危險，與SwimNet一致
• 分區監控——按攝像機區域配置偵測規則
• 符合個人資料（私隱）條例——不儲存或傳輸可識別影像