Image similarity also called reverse image search has use cases far beyond just e-commerce. Explore large unlabeled digital image archives by display similar images. Clean up machine learning datasets by removing duplicated training images that might otherwise be used for training and testing.
Let your users find visually similar photos or products, fast and automatically.
Remove duplications, cleanup similar images, and products accidentally added multiple times.
It simply works on images. As long as you work with images you have a use case for it.
One of our clients is using the image similarity search on a rapidly growing digital archive of around 200000 images. The digital archive consists of many diverse types of images, we are using machine learning to return meaningful similar images in the proper context.
Define a schedule and continuously update your image similarity index.
Always understand your data and the status of your similarity index.
Monitor the history of your similarity index.
Frequently asked questions.
Let us help answer the most common questions you might have.
Which file types are supported
We currently support png, jpg, gif (not animated), gif (animated, will take first frame), bmp and tiff. If you need other file types please let us.
How many results are returned?
You can define how many results you want to get from our API. You can get up to a couple of thousand results for a single request (though we recommend to use just the first couple of results)
Can I see how similar the images are?
Yes. We return the distance, the smaller the distance the higher the similarity.
How can I import images?
You can import your images using our API. We provide integrations for https, Google Cloud Storage and Amazon S3.
Is there a maximum of images?
Our clients are currently running the Image Similarity Search up to 3 million images. If you have more let us know we run some benachmarks if required.