Complete guide for using the catalytic converter grading system
1-5) in the expanded view for faster gradingThe SCA Grading Confirmation System is a web-based application for reviewing and grading catalytic converter images. It uses session-based authentication to provide secure, temporary access to grading interfaces. ML models (King and Dente) provide automated predictions that graders can reference during manual review.
The grading page has four main areas:
Shows your session ID, category, and a progress bar. Tap the chevron to expand details including ML model accuracy scores and prediction summaries.
Thumbnail cards arranged in a responsive grid. Each card shows the image, grade buttons, and any existing grade. Tap the image to open it in a larger modal view.
Expanded view with zoom, pan, and rotate controls. Navigate between images with arrow keys. Grade with number keys. Close with Escape.
Always-visible bar showing graded count, submit button, and actions like Clear, Regrade, and Approve.
Toggle between light and dark themes using the sun/moon icon in the navigation bar. Your preference is saved and persists across sessions. The system also respects your OS-level color scheme preference.
The system supports multiple grading categories. The buttons shown during grading depend on the session's category setting.
Used for IR Grade sessions. LOT sessions show ML prediction data (King and Dente models) alongside each image for reference.
Keyboard shortcuts are available when the image modal is open, allowing fast grading without using the mouse.
| Key | Action |
|---|---|
1 - 5 | Select grade (corresponds to the grade button order) |
← → | Navigate to previous / next image |
Esc | Close the image modal |
IR LOT sessions are created by searching for a 4-character LOT code. The system fetches all units in that LOT and creates a grading session with ML predictions from King and Dente models.
The Reports page provides a comprehensive view of all graded images across sessions, with ML predictions and grader-level detail.
Three KPI cards at the top show: Total Images (all graded images), Unique Graders (number of people who have graded), and Grade Categories (distinct grade values used).
The table shows each image with its unit code, King and Dente ML predictions with confidence scores, Master Tag (consensus grade), grader count, and individual grader columns. Click column headers to sort.
Click any image thumbnail to open a detail modal with zoom/pan/rotate controls, full prediction data, and grader details.
Click “Export Excel” to download the report as a CSV file. The export includes all columns visible in the table. When filters or sorting are active, the export reflects the filtered/sorted data.
Authorized graders can change grades directly from the reports page by clicking on a grade badge. A popover appears with all available grade options. Changes are saved immediately.
Each major section of the application includes an interactive guided tour that highlights key features and explains how to use them.
# Create Grading Pad
POST /api/ir-grading-pad
Content-Type: application/json
{
"user_identifier": "user@example.com",
"expires_in_days": 7,
"category": "am",
"metadata": {
"created_by": "admin-interface"
}
}
# Validate Session
GET /api/ir-grading-pad/{sessionId}
# Update Session Status
POST /api/ir-grading-pad/{sessionId}/status
Content-Type: application/json
{
"status": "approved",
"password": "..."
}# Submit Batch Grades
POST /api/ir-grading-pad/{sessionId}/grade
Content-Type: application/json
{
"grades": [
{
"image_name": "catalytic_001.jpg",
"box_uid": "12345",
"grade": "AM",
"catalytic_id": 1
}
]
}# Fetch Report Page GET /api/ir-grading-pad/report?page=1&limit=50 # Response includes: # - total_images, total_pages, has_more # - images[] with unit_code, king/dente predictions, # master_tag, graders[], image_url
If you need assistance:
Two ML models provide automated classification predictions:
Predictions are polled every 15 seconds during grading sessions and include category labels with confidence percentages.