En | Ru

Neural network image search for ReRooms

Computer vision

System of search which is based on computer vision

Solution

Task

Ability to search for images of interior elements (furniture, accessories, floor coverings) in the ReRooms database. Interior designers work with this base.

Client

ReRooms is an online furniture store, finishing materials and household goods where you can also order a design project and turnkey repairs.

What does it look like from the inside

01/

Dataset of images which based on the material provided by client.

System of search based on MAGE platform

API for sending data of updates

Development stages

02/

Preparing data: indexing and describing images, deleting duplicates

Stage 1

Testing computer model the other
part of the images

Stage 4

Educating a neural network on one part of the images

Stage 3

API development for fast transferring database updates to the system

Stage 5

Selection of computer models, construction of hypotheses

Stage 2

Examples of how the system works

03/

Period of development

04/

1 month

July-August 2021

When we copy the client's database, ReRooms updated and we received extra part of data which cannot be deleted

Problem

All dataset of images had to be downloaded again. To do it we have to wait sunday's night, when ReRooms does not change

Solution

We have improved the API so that the search system quickly responds to changes in the database and re-indexes old and new products.

Solution

The assortment of the client's goods changes daily: for example, some sofas were sold out, others were brought.

Problem

Difficulties of project

05/

Findings

06/

When we work with dataset of images we found a lot of duplicates.
Due to this we avoid a lot of errors during the exploitation, at the same time client actualized info about their own range