AI AOI for PCBA AI 智慧監控 關於 DaoAI 預約專屬時段

讓 AI 執行 AOI 編程,
決定怎麼檢測
讓你的監視器跑
你自己的 AI

AI AOI that programs itself.
Surveillance that learns your site.

AI Agent 接手 AOI 編程、判讀與調參,一塊參考板,5 分鐘上線。
用你自己的畫面訓練 AI 模型,不依賴公版,不上傳雲端。

AI Agent takes over AOI programming and tuning — one Golden Board, live in 5 min.
Train your AI models with your own footage — no generic weights, no cloud upload.

Date
2026.06.02–05展期四天
Booth
Canada Pavilion S0506加拿大館
Venue
TaiNEX 2台北南港展覽館 2 館
At the booth

展位上,我們為您展示

Two things you'll actually see at the booth

/ AI AOI for PCBA
一塊新板子,5 分鐘上線
A new board, online in 5 minutes
把板子放上去——AI Agent 自己辨識元件、畫 ROI、算閾值。全程沒有 CAD、沒有元件庫、沒有工程師編程。
Drop a board onto the rig. The Agent identifies components, draws ROIs, sets thresholds — all of it. No CAD, no library, no engineer in the loop.
/ SKYVISION
20 分鐘搭一個 PPE 稽核任務
Build a PPE auditor in 20 minutes
SkyVision 把四個區塊拖一拖——模型、篩選、邏輯、視覺化——就能組出 PPE 配戴、漏工序、或危險區域闖入的稽核任務。0 代碼。
Drag four blocks — Model · Filter · Logic · Visualization — into a PPE check, a missed-step detector, or a hazard-zone watch. Zero code.
Manufacturing Intelligence

DaoAI AI AOI for PCBA

視覺檢測智能體 · 把 3 小時的編程,壓進 5 分鐘

智慧畫框 + 自動生成檢測閾值,實機操作
97%
編程時間下降
Programming time ↓
80%
誤報率下降
False call rate ↓
60%
營運成本下降
Operating cost ↓
≥98%
檢出準確率
Detection accuracy

AOI 編程工程師的位置,Agent 接下來

元件辨識、ROI 繪製、閾值計算——這三件原本要花 3 到 5 小時人工的事, AI Agent 從一塊參考板自動跑完。換線時間壓進 5 分鐘,操作員 30 分鐘上手。 沒有 CAD,沒有元件庫,沒有 AOI 編程工程師。

它也搞定規則式 AOI 一直卡關的判讀:依工廠合格標準分類 DIP 焊點、從黑電感與石英振盪器的密集字樣中分離真實缺陷、辨識與板材同色的元件、SOD 與 LED 極性。 多維度特徵分析取代 RGB 規則比對。

「一框多測」讓單一 ROI 同時偵測 5 類缺陷:缺件、錯件、反件、損傷、立碑。 1ms / 元件、45 cm²/s 吞吐、檢出 ≥98%。 Agent 隨作業員回饋持續迭代,誤報率下降 80%,越用越準。

The Agent takes the AOI seat

Component identification, ROI drawing, threshold setting — three things that used to take 3–5 hours of manual work. The Agent runs them from a single Golden Board. Changeover collapses to 5 minutes. Operators ramp up in 30. No CAD, no library, no AOI engineer.

It also closes the cases rule-based AOI keeps missing: DIP solder-joint classification to your accept criteria, separating real defects from dense markings on black inductors and crystal oscillators, recognizing components that match the PCB laminate, SOD / LED polarity. Multi-dimensional features replace RGB matching.

One-box multi-detection catches all 5 defect types in a single ROI — missing, wrong, reversed, damaged, tombstoned. 1ms per component, 45 cm²/s throughput, ≥98% detection. The Agent iterates from operator feedback. False calls keep dropping the more you run it — down 80%.

Multi-dimensional AI inspection

多維 AI 檢測,攻克檢測難題

Multi-dimensional AI inspection that cracks the hardest cases

AI 元件對齊,髒汙也能精準定位
AI alignment — pinpoints components even under contamination
同色干擾檢測,與板材同色照樣抓
Same-color interference — finds parts that blend into the board
DIP 焊點分類,依你的工藝標準
DIP solder classification — trained to your accept criteria
黑電感、晶振字元檢測,從根本除誤報
Black inductors & oscillators — separates markings from real defects
Surveillance Intelligence

DaoAI World SkyVision

1 小時構建專屬視覺監控 AI

SkyVision 監控介面

廠務主管和 EHS 團隊,不靠 AI 工程團隊,也能在一小時內把自己的視覺 AI 上線。 七種自訓練模型(監督分割、分類、例項分割、關鍵點檢測、混合、目標檢測、文字識別) 加上 DaoAI 為 SOP 監測、產線堵塞、PPE 配戴、危險區域闖入、人員離崗等場景優化過的工作流—— 依任務選模型、設定偵測邏輯(時間/區域)、拖視覺化模組組成監測任務。

Plant managers and EHS teams build their own visual AI in under an hour — no AI engineering team. Seven self-trainable models (supervised segmentation, classification, instance segmentation, keypoint, hybrid, object detection, OCR) plus DaoAI-optimized workflows for SOP audits, line congestion, PPE compliance, hazard-zone intrusion, and absentee detection. Pick a model, set the logic (time/zone), drop the visualization. Done.

預約 SkyVision demo / Book a demo
01
七種自訓練模型,還有 SOP 偵測等 DaoAI 優化場景
Seven self-trainable models, plus DaoAI-optimized scenarios like SOP detection
監督分割、分類、例項分割、關鍵點檢測、混合、目標檢測、文字識別(OCR)——七種自訓練模型,加上 DaoAI 預訓練模型(無需訓練即可使用)。上傳自有影像、智慧輔助標註、完成自有模型。
Supervised segmentation, classification, instance segmentation, keypoint, hybrid, object detection, OCR — seven self-trainable models, plus DaoAI pre-trained models (ready to use, no training needed). Upload your footage, label with AI-assisted tools, ship your model.
02
拖區塊,不寫程式
Drag blocks, not code
模型、篩選、邏輯、視覺化——四個區塊拼出任務。PPE 配戴、漏工序、危險區域闖入、設備狀態。0 行程式,也不被單一演算法綁住。
Model · Filter · Logic · Visualization — four blocks build a task. PPE compliance, missed-step detection, hazard-zone intrusion, equipment status. Zero code. No algorithm lock-in.
03
全地端,影像不離廠
Fully on-premise. Video stays put.
訓練、推論、儲存全在廠內。攝影機、模型部署、即時告警、GPU 調度、事件回溯,一個介面管完。影像不出廠區內網。
Training, inference, storage — all inside the plant. Cameras, deployment, alerts, GPU orchestration, event traceability in one interface. Video never leaves your network.
1-on-1 時段有限 · Slots are limited

預約 30 分鐘專屬時段

Book a 30-minute private slot

填完我們會在一個工作日內回信,確認 6/2–5 之間的時段,附上展位地圖與你那場的議程。
We'll reply within one business day to confirm your slot during June 2–5 with a booth map and a short agenda.

不填表單也行,直接寫信:business@daoai.com ・Or just email us.