This guide walks users through how to identify and understand image validation failures flagged by AI when a workflow has been run. Failures can be viewed in the project show page and in the Photo Viewer. If this feature is turned off, users will not see any AI Image Analysis data, even if AI Image Analysis has already been run on their projects. Users can also provide feedback on AI Results and View Feedback.
Step 1: Identify Fields with Failed Validations
- Navigate to the Project Show page.
- Review the fields displayed on the page.
How indicators work:
- If a field contains at least one validation failure and has an associated photo, a red dot indicator will appear in the upper-right corner of that field.
- Hover your cursor over the red dot to view additional details.
Tooltip message:
- When hovering over the indicator, you will see:
“Review photos for validation failures.”
This indicator helps you quickly locate and review fields that may need correction.
🟡 Results That Need Manual Review
In addition to passed and failed results, a validation can come back as Needs Manual Review. This means the AI could not confidently determine a pass or fail. These photos display a yellow indicator (rather than the red failure indicator), the validation message header appears in yellow, and the message explains why the result was inconclusive.
- Photos that need manual review are not included when filtering by Failed Validations Only
- They can be resolved after a manual check — see Needs Manual Review: Inconclusive AI Validation Results (link the new article once published)
[SCREENSHOT 1.4 — PENDING TEST] Red failed indicator next to a yellow needs-manual-review indicator for contrast. Place inside this new section.
📝 Example Use Case (with Image Validation)
Imagine you're reviewing a roof inspection project. The field for “Roof Condition” is configured with the following requirements:
- ✅ Image Validation: Roof must be present and detectable in the image
During review:
You notice two thumbnail images with a red warning icon:
Clicking one of the images opens it in the project modal viewer, where the right-hand metadata panel displays an error message:
Image Validation Failed The object 'roof' was not found in the image.
This indicates that the AI-driven image validation could not identify a roof in the image, possibly due to poor angle, obstruction, or the photo being unrelated.
This helps you quickly flag or replace invalid content, ensuring your inspection reports maintain accuracy and completeness.
👀 Step 2: Reviewing Failed Photos in the Photo Viewer
When a user opens the Photo Viewer in a completed project:
🔹 Thumbnail Indicators
- Any photo that fails validation will display a red triangle icon overlay on its thumbnail.
- This helps you quickly spot which images need attention without opening each one.
🖼 Step 3: Viewing Details of Image Analysis Results
Clicking a photo or video thumbnail opens it in full view.
📋 Right Sidebar: AI Metadata
- In the right-hand column, alongside other AI metadata, you’ll see a section labeled:
- This section lists:
- Category
- Caption
- Tags
- Description
- Validations: Pass (Green Text) & Fail (Red Text)
- Extracted Text
Example of a Failed Validation:
⚙️ Step 4: Give Feedback on AI Results
- Clicking "Give Feedback on AI Results" opens a full-size modal where you can submit your input.
📝 How to Submit Feedback
-
Select a Category from the dropdown menu:
- 🏷 Tags
- 🔎 Extracted Text
- ✅ Validations
- 📌 Other
2. Enter Your Comments (required)
- Provide details about the issue or suggestion.
- Be as specific as possible to help improve AI performance.
3. Choose one of the following:
- 💾 Save – Submit your feedback.
- ❌ Cancel – Close the modal without submitting.
💡 Why Submit Feedback?
Your feedback helps improve AI accuracy and ensures better results across future projects.
💡 Tips
- Use the validation icons as a triage tool before deeper photo review.
- You can still manually override or add comments to explain or justify failed photos, if applicable.
- Make sure your project teams understand what each validation rule is checking.
Comments