1. 設定鏡頭 1. Camera Settings
2. AI 辨識分析結果 2. Real-time Analysis
點擊「開始即時辨識」以開啟辨識分析。 Click "Start Predict" to initiate real-time computer vision.
AI 預訓練模型辨識器 AI Pre-trained Vision Lab
🤖 選擇預訓練模型：
🤖 Select Pre-trained Model:
💡 模型說明與預測清單 💡 Model Features & Guides
🎯 COCO-SSD 模式 (Common Objects in Context - Single Shot MultiBox Detector)（Microsoft & Google 開源）： 🎯 COCO-SSD Mode (Common Objects in Context - Single Shot MultiBox Detector) (Microsoft & Google Open-Source):
- 物件檢測 (Object Detection)：能在畫面中同時找出多個物體，並即時畫出追蹤框 (Bounding Boxes)。
- 80 種日常物品：包含人、手機、手錶、杯子、椅子、剪刀、書本等日常物品。
- 🔗 物體清單：點此查看 COCO 80 種可辨識物品清單 ↗
- Object Detection: Locates multiple objects in frame simultaneously and draws Bounding Boxes in real time.
- 80 Common Classes: Detects everyday items including people, phones, chairs, scissors, etc.
- 🔗 Class List: View COCO 80 Categories Reference ↗
🔒 隱私與備份機制:
🔒 Privacy & Fallback Protection:
- 零影像記錄：所有 AI 影像分析均在您的個人裝置（電腦/手機）本地即時運行，絕不記錄或傳輸至雲端。
- 雙重備份防護：若 CDN 載入超過 5 秒或受阻，系統會自動轉載伺服器本地備份模型，確保課堂體驗順暢。
- Zero Cloud Storage: Inference runs strictly in local device memory with no video recording or data uploads.
- Fallback Protection: Automatically falls back to local server assets if CDN load exceeds 5 seconds.
⏳ 正在載入系統與預設模型...
⏳ Loading system and default model...