Similar images

Want to have a more like this search for visual content? Use visual search to find similar images in your collection. 

Solr integration

pixolution flow is a plugin for Apache Solr. Both systems run seamlessly together. This makes Solr not only a great search engine for textual content but also for visual content. 

Query by URL

Use sample images from the web or your local hard drive to find identical or similar images in your collection.

Multi-color search

Enable your users to find images that suit specific color requirements and match the color scheme of a website or brochure. Users can define one or several colors and can weight specific colors to increase their impact on the ranking.

Relevance weighting

Influence the scoring mechanism and weighting between the importance of textual and visual relevance.

Smart filter

The smart filter is a pre-scoring technique that filters out images irrelevant to the query image. Depending on the filter level it can drastically reduce the amount of images that have to be scored and improves search performance tremendously while keeping quality loss to a minimum.

Cloud descriptor lookup

When running in cloud mode, pixolution flow will automatically look up and fetch all relevant information such as image descriptors, filters, or metadata from the server storing that data in order to process the search query. Just query any server in the cloud; pixolution flow does the rest.

Duplicate detection

Detect duplicate visual content without metadata. Instead, the pixel data is analyzed to detect image duplicates with different sizes, file names, file formats, encoding quality, or visual modifications. Deduplicate your existing collection and continuously check new images before indexing them.

Lightning fast

Our image search is lightning fast due to small image descriptors, smart filtering, and distributed search capabilities. And the best thing: it runs on commodity hardware.

Auto context

If your sample image contains indexed metadata, pixolution flow can automatically extend the query to include the keywords associated with the sample image. This strengthens the semantic context of the visual search.

Image descriptor generation

Do you preprocess your data before indexing into Solr? Using our HTTP API, Java API, or via command line interface you can generate image descriptors for images, store them in a database, and simply index the already available image descriptors.

Text space filter

Do your users need space for text or logos in the images they are searching for? With our text space filter users are able to define blank areas. This is perfect for finding images that are suitable for further creative editing, such as book covers, brochures, and posters.

Combined text and image search

Easily combine visual search criteria with other traditional search criteria such as category filters, keywords, date ranges etc. in one query.  

Image tagging

Unlike other image tagging systems, pixolution flow uses your previously tagged images to suggest tags for new images. This approach ensures a consistent wording style, a domain-specific language, and the tagging quality will improve as your media database grows.

HTML response format

Relevance tuning is much easier when you actually see the result images rather than a JSON response. We’ve added an HTML response writer that enables developers to inspect the search results visually without implementing a UI first. Send a GET request and see the HTML formatted response right in your browser.

Define minimum relevancy

Sometimes the most similar images in your collection may be still irrelevant to your query. You can define a threshold so only images with a minimum relevance are returned.

Universal connector

Not using Solr as your search engine? Use pixolution flow in combination with your existing search engine. Hand over the search results to pixolution flow and search visually within this subset. You can retain your already implemented search logic and filter options.

Automatic schema configuration

No need to configure fields and fieldtypes. pixolution flow does it for you and supports the managed schema mode. But you can also configure the schema manually if you prefer.

Memory efficient

We’ve developed extremely small image descriptors to represent the visual information of an image in less than 150 bytes per image. By comparison, the average file size of a thumbnail image is about 20,000 bytes. This minimizes our memory footprint and I/O.

Runtime configurator

Most options can be set and changed at runtime on a per-query basis. This enables you to quickly test and adjust the configuration to your needs without restarting. Set the optimal configurations for different use cases.

Runtime stats

pixolution flow creates usage information such as download and search speed, or how often it was accessed, and provides these stats via Solr API. This is useful for developers who want to know what’s going on behind the scenes.


With pixolution flow, you’re prepared to grow. We support the Solr cloud mode with replication and sharding. Split and replicate your collection across several servers to handle massive parallel user requests and huge collection sizes. Searching in millions of images is easy.


When using external images as search input, the images are downloaded and analyzed on the fly. pixolution flow caches the calculated image descriptors and their associated URLs. If the same resource is used again, the image descriptor will be fetched from the cache.