How to Verify Image Content Analysis is Active in Vertex AI Search for Commerce

Google Cloud Build with AI AI Solutions
1

Hi everyone,

I’m currently using Vertex AI Search for Commerce and I’m trying to ensure that the visual content of our product images (e.g., colors, patterns, styles) is being used to enrich our search index and improve product discovery.

My main concern is that visual attribute search is not working, which leads me to believe that the image content isn’t being analyzed or indexed correctly.

For example, we have a product titled “Snoopy With Love”.

  • A search for the exact title “Snoopy With Love” correctly returns the product, so we know it is indexed.

  • The product’s image clearly shows a red dog figure holding a heart. However, a search for the visual description “red dog with heart” returns zero results.

This test strongly suggests that Vertex AI Search is only matching against the text data (title, description, etc.) and is not leveraging the visual information from the product image.

This leads to my main questions:

  1. Is there a specific setting or requirement we might be missing to activate this image analysis feature?

  2. How can we conclusively verify that visual attributes from our images are being processed and are influencing our search results?

  3. Is there any tool, log, or API endpoint that would allow us to see the AI-generated attributes that the system has extracted from an image for a specific product?

  4. Is there a typical delay or processing time after catalog ingestion before image-based attributes become searchable?

Any clarification on how this integrated feature works and how to troubleshoot why it might not be working in our case would be greatly appreciated.

Thank you!