Search for a similar image would return results that visually relate to the selected design. If you work with the image tags indexing function with Clarifai AI, you can search for visually similar images.
How does it work?
- Clarifai identifies similar images using Clarifai AI models trained on large datasets.
- It analyzes images by extracting visual details like colors, shapes, textures, and patterns.
- Each image is processed into a numerical form called a feature vector, representing its unique visual characteristics.
- To find similar images, Clarifai compares these vectors using methods like cosine similarity or Euclidean distance. Images with closer vectors are considered more similar. For instance, if you select a design with squares, borders, and brown color, then it is going to search for "similar" designs that contain those characteristics.
- Results are ranked by similarity, and on neoCatalog advanced settings, we can adjust the value of how sensitive the matching process is. This value can be adjusted in the advanced preferences.
Once you have made the setup with Clarifai, the search for similar images is done by a button in the header bar on the home design page:
Step-by-Step
1. Go to your free Clarifai account and create a new application following the setup with Clarifai.
2. Go neoCatalog Categorization, configure the part of Tags with required configuration data, and save.
3. Once the administration part for tags is configured, you have to update image tags for the image search of all designs. This can take some minutes. In Clarifai App Explorer you can view tagged designs.
4. Restart the neoCatalog service (required).
5. Start searching similar images with the button next to the quick-search field.
6. You will get the highest precision in the first result and then be ordered by the least precision. If you look for higher precision, the administrator can create advanced preferences and define the precision value.
Example:
Related articles:
How to tag designs with Clarifai AI™