Image Similarity Search

Do not restrict your users to search by keywords & attributes. Give them visually similar items, quickly and automatically. Accurate product recommendations keep users happier and increase your conversion rate.

Visual Search in Fashion

Fashion e-shops and comparator sites often display related products on each item page. Ximilar provides similarity search solutions enabling them to show these product recommendations, i.e., visually similar items, automatically due to image similarity API.

How does image similarity search work? Your collection is uploaded to Ximilar cloud, where our neural networks process and analyze your images. An image similarity API call then provides the most similar images to a product in a particular category. We constantly update our visual similarity search for fashion. Our powerful visual AI can search millions of items in a fraction of a second.

Given Product

Capabilities

10s of Millions of Product Images
Results available in 100s of milliseconds

Visually Similar Products

85%

79%

77%

75%

71%

66%

63%

62%

59%

56%

53%

Try it yourself: See demo

Similar Photos & Search by Image

E-shops, stock photo agencies and big data collections typically browse their collections with tags, labels and other attributes. Ximilar, on the other hand, focuses specifically on the visual data and analyzes every image in detail, using visual AI and machine learning. With visual search, your customers can see product recommendations they would never get by attribute filters.

This way, you can also implement reverse image search (also search by image). In this case, the query is an uploaded external image, used for a reverse search for the same or similar images in your collection. Our best image similarity models are built and tuned especially for fashion products and generic images, and we add new ones based on the customers’ requests.

Uploaded Photo

Capabilities

10s of Millions of Images
Results available in 100s of milliseconds

Visually Similar Imagery

85%

79%

66%

71%

79%

77%

85%

67%

68%

83%

74%

65%

Try it yourself: See demo

How is the Image Similarity Analyzed?

Ximilar measures the mutual similarity of images from several viewpoints – semantic content (what is on the image), color impression, keywords & tags, etc. For specific areas, like fashion, our similarity model learns from hundreds of thousands of pairs of similar fashion products. We can also create a specific similarity function or a custom image search AI for you – just talk to us about your use case.

We are Experts
Ready to Help You Out

Machine Learning is our passion. We know all the ins & outs of technologies that might help you level-up your business using full or partial automation. Contact us and let's speak about what we can do for you.

Technology Stack

We use state of the art neural network models & machine learning techniques. Constantly trying to improve the technology so you always have the best quality available. Each model has millions of parameters which can be processed by CPU or GPU. Our intelligent algorithm picks from several models & uses the best performing.

API Documentation

We take care of the complexity behind and wrap it in a few lines of code.

full documentation

curl -H "Content-Type: application/json" -H "authorization: Token __API_TOKEN__" -H "search-collection: my_collection" https://api.ximilar.com/similarity/photos/v2/visualKNN -d '{"query_record": {"_url": "https://bit.ly/2IymQJv" }, "k": 10 }'

import requests 
import json 
 
url = 'https://api.ximilar.com/similarity/photos/v2/visualKNN/' 
headers = { 
    'Content-Type': 'application/json',
    'Authorization': "Token __API_TOKEN__", 
    'search-collection': "my_collection"
} 

 
data = { 
    'query_record': {'_url': __IMAGE_URL__ },
    'k': 10
} 
 
response = requests.post(endpoint, headers=headers, data=json.dumps(data)) 
if response.raise_for_status(): 
    print(json.dumps(response.json(), indent=2)) 
else: 
    print('Error posting API: ' + response.text)

$curl_handle = curl_init("https://api.ximilar.com/similarity/photos/v2/visualKNN");

$data = [
    'query_record' => [ '_url' => 'https://bit.ly/2IymQJv' ],
    'k' => 10
];

curl_setopt($curl_handle, CURLOPT_CUSTOMREQUEST, "POST");
curl_setopt($curl_handle, CURLOPT_POSTFIELDS, json_encode($data));
curl_setopt($curl_handle, CURLOPT_RETURNTRANSFER, true);
curl_setopt($curl_handle, CURLOPT_FAILONERROR, true);
curl_setopt($curl_handle, CURLOPT_HTTPHEADER, array(
        "Content-Type: application/json",
        "Authorization: Token __API_TOKEN__",
        "search-collection: my_collection",
        "cache-control: no-cache",)
);

$response = curl_exec($curl_handle);
$error_msg = curl_error($curl_handle);

if ($error_msg) { // Handle error
    print_r($error_msg);
} else { // Handle response
    print_r($response);
}
curl_close ($curl_handle);