Step D: Visuals Comparison with & without Measures
In the Power BI Desktop screen
1. Optional: you can add the “Simple Image” custom Visual so you can view a larger image instead of the thumbnail sized image in the matrix visual.
 |
 |
- Populate the Image URL Field of the visual with the ImageURL field from the Images table.
2. Add a Matrix visual and set its field properties
- Notice when the Image is collapsed in the Matrix visual, then the Tag Value is incorrect.
|
The Tag value for 99.91% should be table and not coffee. |
 |
- To correct this, we have two options when the row is collapsed.
Tag column shows the correct Tag value for the corresponding maximum Confidence %.
OR
Tag Column shows comma separated value of all Tag values, and Confidence % is blank.
4. We will need to use the DAX measures to correct this behavior, Create another matrix visual with the [Tag Measure] & [Confidence Measure] that we created in the previous blog post and rename the matrix fieldnames to Tag and Confidence %.
5. If you would like the Comma Separated value of all the tags when collapsed with a blank confidence value, then alternate the commented-out section of the [Tag Measure] in the else portion of the IF statement.
Conclusion:
In these series of blog posts, we have seen
- Smart data extraction algorithm of the Web by Example feature in Power BI.
- Image Tag extraction using AI Vision Insights in Power BI
- Using DAX Measures to control the UI behavior of the Matrix Visual.
The Image Tag extraction feature allows us to extract the object names from an image, and then perform Power BI analytics (filters, grouping, …) on the extracted text.
Image License:
The blurred laptop image in the screenshot was sourced from https://unsplash.com/photos/yC-Yzbqy7PY/download
At the time of this the image license terms is listed below.
