Image-Based Search – How to Find Items Using a Photo

What is the image search?

The image search on Galeria Savaria is an innovative feature that allows you to find identical or similar items based on a photo on the online marketplace.

You don’t need to know the exact name, type, manufacturer, or artist – if you have a photo, that’s enough. The system will automatically search for the most relevant matches based on the image.

Where is it available?

The image search is available on the Items page and in the Archive Catalogue, in the latter case, it is available to subscribed users (until further notice, it is also accessible without a subscription).
It is still under development and continuously learning.

• Go to the Items page or Archive Catalogue page
• Click the camera icon in the search bar.
• Upload or drag your image into the box.

How does it work?

The system compares the item in your uploaded image with over one and a half million sold and archived items in the Galeria Savaria database.
It ranks the most similar items based on various visual characteristics, such as shape, color, pattern, and finer details.

The results page of the image search remains available for 24 hours.

Important information

🧠 The search is powered by artificial intelligence – and it's still learning

The image search is powered by an AI-driven system that ranks results based on visual similarity, but it is not perfect yet.

Although it can already recognize identical or stylistically similar items, the search may sometimes consider visual elements that aren’t truly relevant (e.g. background, lighting, photo angle).

This is why it’s important that the system is trainable – so it can learn over time which details truly matter and which do not. For this training process, we invite the help of our users – including you. You can read more in a dedicated article.

Would you like to help improve the image search?

If you’d like the search to become even more accurate at finding matching items, you have the option to participate in training the system.

👉 Learn more here:
📘 Training Mode – How You Can Help Teach the Image Search