Session Feedback
The Session Feedback view allows you to provide real-time feedback on PhotoRater AI detections during rating sessions. Your input helps improve AI accuracy and provides visibility into detection performance.
Skill Levels: The detail level of AI feedback you see is controlled by your assigned skill level (Trainee, Intermediate, or Advanced). See Skill Levels to learn how feedback adapts to your experience level.
What Is Session Feedback?
Session Feedback is a quality assurance system that lets photographers indicate whether AI detections were correct or incorrect. This serves two purposes:
- Photographer Review: Review your feedback submissions during the current session
- Quality Assurance: See visual accuracy metrics for each detection type to understand AI performance
Non-Intrusive Design: Feedback is completely optional and doesn't affect your rating workflow. Photos are rated normally whether you provide feedback or not.
Accessing Feedback View
- Look for the Feedback icon in the sidebar (purple thumbs-up icon)
- Click to open the Session Feedback view
- View appears with accuracy charts and images with feedback
The Feedback view is available during active rating sessions and resets when the session ends.
[IMAGE: Feedback View Overview]
Feedback view showing accuracy charts and image cards
Providing Feedback
Feedback is submitted directly from the Image Detail view while rating photos.
Feedback Buttons
When viewing individual photo details, you'll see thumbs up/down buttons below each detection result:
- Thumbs Up (Purple): Click if the PhotoRater AI detection was correct
- Thumbs Down (Red): Click if the PhotoRater AI detection was incorrect
[IMAGE: Feedback Buttons in Image Detail]
Thumbs up/down buttons below detection results
Button States
| State | Appearance | Description |
|---|---|---|
| Unselected | Gray background | No feedback submitted yet |
| Hover | Purple tint on gray | Preview of selection |
| Correct Selected | Purple background (#9333ea) | Marked as correct, button disabled |
| Incorrect Selected | Red background (#ef4444) | Marked as incorrect, button disabled |
One Submission Per Detection: Once you click thumbs up or down, the buttons become disabled for that detection. Currently, you cannot change your feedback after submission.
Detection Types
You can provide feedback on these five PhotoRater AI detections:
| Detection Type | What It Detects | Example Feedback Scenario |
|---|---|---|
| Pose Detection | Head angle, shoulder angle, body positioning | AI says pose is good, but head is actually tilted too far |
| Eyes Detection | Open, closed, or blinking eyes | AI missed that subject's eyes were closed |
| Glare Detection | Glasses glare and reflections | AI detected glare that isn't actually problematic |
| Smile Detection | Facial expression and smile presence | AI correctly identified natural smile |
| Blur Detection | Sharpness and focus quality | AI flagged blur, but image is acceptably sharp |
Viewing Feedback Summary
The Feedback view displays your session feedback in two main sections:
Accuracy by Detection Type
Donut charts visualize feedback accuracy for each detection type with a purple/red color scheme:
- Purple (#9333ea): Correct feedback (AI got it right)
- Red (#ef4444): Incorrect feedback (AI got it wrong)
Chart Layout
Charts are displayed in a responsive grid:
- Mobile: 2 columns
- Tablet: 3 columns
- Desktop: 5 columns (one per detection type)
[IMAGE: Donut Charts Grid]
All 5 detection types with accuracy percentages
Chart Components
Each donut chart displays:
- Center Percentage: Overall accuracy for that detection type
- Purple Arc: Proportion of correct detections
- Red Arc: Proportion of incorrect detections
- Stats Below:
- Correct count (purple text)
- Incorrect count (red text)
- Total count (gray text)
Images with Feedback
Below the accuracy charts, a grid displays all photos that received feedback during the current session.
Image Card Components
| Component | Description |
|---|---|
| Image Preview | Square thumbnail with rounded corners |
| File Name | Photo file name (truncated if long) |
| Rating Display | Quality rating (1-10 scale) with orange stars |
| Feedback Badges | Shows each detection type that received feedback with thumbs up/down indicator |
| Summary Stats | Total correct (purple) and incorrect (red) count for this photo |
[IMAGE: Image Card with Feedback]
Photo card showing feedback badges and stats
Feedback Badge Color Coding
- Purple Background: Detection marked as correct
- Red Background: Detection marked as incorrect
- Thumbs Up Icon: Correct feedback
- Thumbs Down Icon: Incorrect feedback
Each badge also shows the ML confidence percentage (small gray text).
Feedback Lifecycle
Complete Workflow
- Start Session: Log in with your PIN
- Rate Photos: Use Quick Rate or Image Detail views
- Provide Feedback: Click thumbs up/down on detection results
- Review Feedback: Navigate to Feedback view to see session summary
- End Session: Session ends, feedback data is stored
When Feedback Resets
Feedback is session-scoped and resets in these situations:
- You end your current session
- You log out of the application
- You start a new session with a different PIN
Real-Time Updates: Feedback appears instantly in the Feedback view after you click thumbs up/down. No page refresh needed.
Empty State
If you haven't provided any feedback yet in the current session, you'll see:
No Feedback Yet
Use the thumbs up/down buttons on image details to provide feedback on PhotoRater AI detections. Your feedback will appear here and help improve detection accuracy.
Understanding Accuracy
How Accuracy Is Calculated
Accuracy percentage is calculated per detection type:
Accuracy = (Correct Feedback Count / Total Feedback Count) × 100
Example:
- Glare Detection: 15 correct, 3 incorrect
- Total: 18 feedback submissions
- Accuracy: (15 / 18) × 100 = 83.3%
What Good Accuracy Looks Like
| Accuracy Range | Interpretation |
|---|---|
| 90-100% | Excellent - AI is performing very well |
| 80-89% | Good - AI is reliable with minor issues |
| 70-79% | Fair - AI is mostly accurate but needs improvement |
| <70% | Needs Attention - Review detection thresholds or provide more feedback |
Sample Size Matters: Accuracy is more meaningful with more feedback. 8 out of 10 correct (80%) is less reliable than 80 out of 100 correct (also 80%) due to sample size.
What Feedback Does NOT Affect
It's important to understand that feedback is for improvement tracking only:
- Photo Ratings: Feedback doesn't change the quality score assigned to photos
- Detection Results: Existing detection results aren't modified
- Future Photos: Currently, feedback doesn't automatically adjust thresholds for new photos
- Other Photographers: Your feedback is session-specific and doesn't affect other users
Future Enhancements: In upcoming versions, aggregated feedback will be used to fine-tune AI thresholds and improve model accuracy across all photographers.
Frequently Asked Questions
Where do I provide feedback?
Click the thumbs up/down buttons below each detection result in the Image Detail view.
Why does my feedback look different from other photographers?
AI feedback adapts based on your assigned skill level. Trainee users see detailed descriptions with guidance, Intermediate users see concise 1-2 word descriptions, and Advanced users see icons with penalty values only. See Skill Levels for details.
What happens to my feedback?
Feedback is stored locally during your session and displayed in the Feedback view. In future updates, it will sync to the cloud for centralized analysis.
Can I change my feedback?
Currently, no. Once submitted, feedback buttons are disabled. This may be enhanced in future versions.
Does feedback affect my photos?
No. Feedback is for AI improvement tracking only and doesn't change photo ratings or detections.
What if I don't want to provide feedback?
Feedback is completely optional. You can rate photos normally without using the feedback buttons.
Next Steps
- Rating Workflow - Learn how to rate photos and view detections
- Sessions - Understand session management and PIN authentication
- Dashboard & Analytics - View insights and metrics