🖼️ SPaCial AI OCR Client (Enhanced)

This interface connects to the SPaCial AI OCR API server for advanced dimension detection.

🎨 Color Legend:

  • 🟢 Green: Regular text (horizontal/vertical)
  • 💚 Light Green: Diagonal text (45°)
  • 🔵 Blue: Dimensions (horizontal/vertical)
  • 💙 Bright Blue: Diagonal dimensions (45°)

📐 Orientation Symbols:

  • → Horizontal | ↓ Vertical | ↘ Diagonal (45°) | ↙ Diagonal (135°)

⚙️ OCR Modes:

  • Fast: Quick detection for well-visible text
  • Accurate: Better for small or unclear text
  • Hardcore: Maximum detection (aggressive parameters)
🌐 Select Server

Choose between cloud (HF Spaces) or local server

Click 'Check API Status' to verify server connection

⚙️ OCR Settings
OCR Mode

Hardcore mode uses aggressive detection parameters
0 270

Process an image to see statistics

📝 Extracted Text Details

🎓 Training Data Collection & Corrections

Review, correct, and export OCR results for model training
Process an image to see detected zones

✏️ Zone Corrections

💡 Tip: Enter zone number to edit, or use 'Add New Zone' to create one manually

📐 Adjust Box Coordinates

💾 Export & Share

Export Format:

  • Original zones (before corrections)
  • Corrected zones (with changes tracked)
  • Statistics (validated, fixed, deleted counts)
  • Timestamp and unique image ID

Telegram Integration:

  • Sends JSON data to training channel
  • Includes all corrections and metadata
  • Automatic notification to team

📋 Quick Start Tips
  1. Upload a technical drawing or document image
  2. Select OCR mode (try 'hardcore' for difficult images)
  3. Adjust rotation if image is sideways
  4. Click 'Process Image with OCR'
  5. View results with color-coded bounding boxes

Note: Dimensions (with Ø, ±, tolerances) are automatically highlighted in blue!