Visual Search Services FAQ
About Visual Search
The ready-to-use solutions based on visual search technology are available for testing in both public demo and Ximilar App.
It is also a standard procedure to prepare custom demos. Read our page How we work and contact us. We will prepare a custom demo tailored to your collection.
Go to:Read more:
Visual & similarity search technology is able to analyze the overall visual aesthetic of an image or detected object in an image, independent of the origin of images or any kind of metadata (such as keywords). It understands the concept of similarity according to your subjective perception. That is why it can provide the most relevant results to image queries, whether you look for the exact match or similar items.
Typical Applications
The term visual search usually refers to searching images or products based on an image query, including photos from social media and user-generated content, such as real-life photos from smartphones. We combined this technology with object detection to create a service called Search by Photo, which is used by e-shops for product searching based on end users’ image queries.
The technology is also often used for similarity search, typically for product recommendations in e-commerce. It analyzes features of an image or detected object, such as colour, edges, or patterns, to recommend the most similar alternatives.
Another typical use is image & product matching. Since the visual & similarity search technology can identify duplicates, nearly-identical items or the same items in images with different quality, it can help with creating product galleries, curating them and eliminating unnecessary duplicates.
Read more about commercial applications on our page Visual Product Search.
Other Applications
Even though applications in e-commerce are the most common, visual search provides endless possibilities even in the industrial sector, scientific research or security systems. Read our articles or contact us for details.
Go to:
- Visual Product Search
- Similar Product Recommendations
- Product Similarity – Public demo
- Photo Similarity – Public demo
- Search by Photo
- Search by Photo – Public demo
- Product Matching
- Articles about visual search
- API Documentation
- What are the examples of visual & similarity search solutions built by Ximilar?
- How Visual AI helps e-shops, price comparators, or collectibles’ sites?
- Can Visual AI help with sorting, filtering, and recommendations of images or products on my website?
- How to find duplicates, enrich product galleries, or add the metadata to the new images?
- How do I know if I need a custom visual search solution?
- How many images do we need to train a visual similarity model?
- Can we build a smartphone app to search for specific products?
- Can visual search engine built by Ximilar be deployed on our servers?
Visual Product Search was built for e-commerce. It is useful for finding products with image queries, similar product recommendations, and product matching.
Photo Similarity works with the same technology, but it was trained for generic images, such as stock photos or real-life images.
Fashion Search is a specialized service for fashion e-commerce, which combines visual & similarity search with object detection (Search by Photo) and Fashion Tagging.
Custom Visual Search refers to solutions using visual & similarity search we build from scratch for our customers.
Go to:Read more:
Yes, you can combine these services. Read more on their pages.
Go to:Read more:
Visual search is always custom designed to search your images based on their visual similarity. To use the visual search, your image database needs to be synchronized to your collection in the Ximilar cloud.
During synchronization, each image is analyzed, a representation of this image is created, and then the image file is discarded. That is why your collection does not contain any actual images. It may, however, contain links (URLs) to the original images, if you provided them.
If your image database is dynamic (images are inserted and deleted over time), your collection at Ximilar cloud must be synchronized. This synchronization can be done via API, or Ximilar can set up a regular synchronization.
Go to:Read more:
The visual search by Ximilar was already successfully applied to video footage, and we are looking forward to developing the technology for other types of data, such as 3D models or VR scenes! Contact us to discuss your use case.
Go to:Read more:
The source of the images in your collection is up to you. It can be third-party data (e.g., scraped from public websites or social networks), but it is up to you to collect these images (files or their URLs) and to act in accordance with the copyright law.
Go to:Read more:
Custom Visual Search
We use our platform to build complex visual & similarity search solutions for the fields of e-commerce, fashion or stock photos. Check out our services Visual Product Search, Fashion Search, Image & Product Matching.
When approached by a new customer, we discuss their use case and data and decide whether we will customize one of the existing solutions or build a completely new system. Either way, we will plan a project and deliver a visual search solution tailored to your data.
Go to:
The more training images you have, the better. We recommend having at least 1000 groups or pairs of images in the beginning. We can iteratively help you build a dataset for training a visual similarity model that can be used in visual search technology.
Go to:
Yes, visual search of products based on user-generated content is a typical application of our technology. Read about our service Visual Product Search to learn more. Contact us to discuss your use case.
Go to:
Ximilar built complex solutions for a number of big e-shops and price comparators. For example, we develop visual & similarity search apps, such as Skintory. Read about visual search solutions to learn more.
We also provide a specialized service for fashion websites called Fashion Search. It combines several visual AI solutions, such as object detection, automatic tagging or object detection, to automate all image processing and level up the customer experience in fashion e-commerce.
Go to:
Technology
If the image is already in the collection, then the search is very fast (under 100 ms). If you are searching with an external image, then this can take up to 0.5 seconds. We are able to scale the system to be even faster if needed. Contact us for details.
Go to:
Yes, the system can be easily deployed on your servers if needed. The server should have a standard Intel or AMD CPU with at least 96 GB of RAM. It would be great if the server also contained an NVIDIA GPU card. We are using Docker technologies that help us with the deployment.
Go to:
Collection Synchronization
Our largest collection contains more than 100 million images, and the speed for searching visually similar items is under a few hundred milliseconds.
Go to:
Yes, we are able to do automatic synchronization from your export, and you can choose both the way and frequency (daily, weekly, or monthly) of synchronization.
Go to:
Your export should be accessible via URLs in standard format as JSON, XML or CSV.
Go to:
You can keep your collection updated via API. We provide standard endpoints for inserting, updating or deleting the items from the collection.
You can also provide your collection as an export file, and we will keep it updated for you. You can choose the way and frequency of collection synchronization.
Go to:
Text-to-Image Search
The search itself happens in a matter of milliseconds. This system can handle hundreds of requests per second if needed.
Read more:The text-to-image search eliminates the need for metadata like tags (keywords). It is powered by AI able to understand both language and complex visual characteristics of images.
Once you synchronize your image collection to our cloud, our AI retrieves distinct structured information (embeddings) from your images. We do not store your images, only the extracted data. When a user enters a text query, our AI thoroughly analyzes it and presents the user with the most relevant content from your collection.
Read more:
