pixolution.io - Trial platform released
We're excited to release our new pixolution flow trial platform at pixolution.io. Now you can easily experience the pixolution flow features using your own set of images.
After signing up, you have an empty collection. The import wizard makes data import easy, just upload a file containing your image URLs and metadata. While importing the data you can start testing right away - even if the indexing process is still ongoing.
Of course, the more data you import, the better the search results can be. Enjoy testing our great features!
With the visual search features you can search for certain images in your collection. There are several options to start a search: You can just click the Find similar button of an image, type in a keyword, choose a color or use an external image as search input to search for similar ones in your collection.
A combination of keywords and visual search criteria is also supported.
Detect duplicate visual content without using metadata. Upload an external image or paste an image URL. The pixels are analyzed to detect image duplicates with different sizes, file names, file formats, encoding quality, or visual modifications. A slider lets you determine the detection sensitivity to fit your needs.
Upload an external image or paste an image URL and get keywords for that image. Keyword suggestions are generated based on your previously imported keyword data. Those keywords act as the ground truth. If there is no or not sufficient data, suggestions cannot be made.
The big advantages of this approach: suggestions are always made in your language and if you have a controlled vocabulary it is taken into account since only previously indexed keywords will be suggested. The wording style is consistent throughout your complete image collection and therefore fragmented results through different wording is avoided.
Bringing visual search to end-user is not only a question of technology, but also of the user interface. Everybody understands to type keywords in a searchbar. We've implemented a searchbar that retains the known search paradigma but - at the same time - supports colors and images as search input.
This is the result of three iterations with customer feedback and demonstrate how visual search should be integrated into a search app. The searchbar is very powerful and handles different input types but hides the complexity of handling them. Simplicity is the best way to increase user adoption.
Just test it yourself. If handling visual input types feels as natural as handling text, then we did it right.
More to come
You would also like to test the features via API? We will provide API access in the near future. You will then be able to integrate visual search in your own project.