User Guide

Informational and technical resources related to our products

Product

Visual search is the next generation of search and that is the way users are going to interact with retail platforms in future. Our product can make photos and videos shoppable by identifying products in them and finding similar products from indexed catalogue.

Autocrop

In past, visual search products required inputs from user for cropping the area of interest from input image. Turing visual search eliminates the need by providing autocrop functionality. As soon as image is uploaded, we identify the products and give most relevant results. This eliminates time consuming cropping operations from user end while making the process very fast and accurate.

How it helps users?

The biggest problem with unstructured categories like furniture, jewellery and fashion is that it’s very hard to describe in words. This is the place where traditional text search fails to do its job. Visual search makes any photo and video content shoppable by providing ability to search directly from media and bypassing text search. That means, your users can get any photos from social media like Instagram, Facebook or Pinterest, as well as photos from their camera and search for exact items which they can’t describe in words.

How it helps retailers?

The biggest problem in retail is the content discovery. Every retailer these days have thousands or even millions of product in the catalogue but showing the right product to user at the right time can help increase conversion for retailers.

Visual search perfectly solves this problem by showing users the most relevant content at the time when user has highest intent to buy it. Matching the high intent with prefect search result has proven very beneficial for the retailers. It also adds delight factor for users as well as confidence in platform as he can find the products they are looking for within seconds.

It can also be used to make any images and video shoppable. Have celebrity pics which you want to make shoppable? Have a videos with home and lifestyle products? Everything can be linked to a product and presented to users to buy.

How it works for users?

Users have to just upload a pic from their device. Our autocrop APIs first detect all objects in the pic and shows user identified objects. Once the user selects his desired item, most relevant search results for the product are displayed.

How to integrate?

We provide very simple REST APIs to integrate the product with your platform. The API documentation is available here.

Research


Visually Similar Recommendations

Product

Visually similar recommendations or simply visual recommendations are recommendation system which are capable to show visually similar products to users who are looking at a certain product. An example is provided in screenshot below.

How it helps users?

In most of retail systems, when users are shopping, they are looking for a certain type of product which fits their liking. Visual recommendations helps them drill down to exact same item they are looking for.

The visual recommendation systems are proven to delight users by providing them personalised in-shop like experience online. Due to the system’s capability to bring out the similar looking products, users have their quest for “something like this” satisfied.

How it helps retailers?

The biggest problem in retail is the catalogue discovery. Every retailer these days have thousands or even millions of product in the catalogue but showing the right product to user at the right time can help increase conversion for retailers.

In unstructured categories, these systems are proven to increase CTR and conversion rates because it presents right products to users who have very high intent to purchase those products.

Older methods like “people who view this also viewed” rely on matrix factorisation methods which fail miserably in unstructured categories like sofa, lifestyle, fashion and jewellery.

How to integrate?

We provide very simple REST APIs to integrate the product with your platform. The API documentation is available here.