Improved photo search system for the marketplace
Client
KazanExpress - a trading platform with goods from Russia and China. Offers one day delivery
01/
Background
In 2020, EORA implemented its own MAGE image similarity search system on the KazanExpress website.
Inside the service is a neural network that is able to analyze and find similar products.
The system speeds up the purchase of goods: instead of using filters, the buyer simply uploads a photo.
The customer decided to increase the accuracy of the search and gave us a new database of images: from stores and from users.
Solution
Task
Increase the quality of the search for similar images
Post training of the neural network on new photos
Speed up the service amid growing number of requests
Optimization of service performance
How it works
02/
The customer sees a list of products with price tags and descriptions
Conversion from browsing to buying increases
The neural network accesses the database of the website and selects similar images
Buyer uploads a photo to search
03/
Quote
Alexey Guchko
«We did the project fast because we worked with an off-the-shelf product we know well: MAGE. We improved the search quality and also created a post-training module. It allows data to be indexed every time the image database is updated.»
Project Manager
Development Stages
04/
Step 01
Data markup: selection of relevant images
Step 02
Training a neural network on new photos
Step 03
API preparation to launch the model
Numbers
05/
minimum throughput of our computer model when running on a server
This is how much the search accuracy has increased after post training the model
10 queries per second
X2 times
Project challenges
Problem
Solution
In the new database, there were many not relevant photos received from users. The search may have included a photo of the packaging, but not of the product itself. Such images were not suitable for training.
We conducted several experiments and selected the training method that gives the highest search accuracy. To post train and test the computer model, we used photos of the seller (store).
06/
Development timeline
07/
15 days
September 2021
Interesting fact
Kazan Express has its own data science team. It creates all the product features - with the exception of tools based on computer vision technology. Those are handled by EORA
Project Team
08/
Vlad Vinogradov
Teamlead
Alexey Guchko
Project Manager
Vyacheslav Schultz
Data scientist
Ivan Izmailov
Backend developer
Technologies
Search for similar images for online stores
Mage
We used our own service
09/
Perspectives
10/
MAGE allows for fine-tuning to the needs of the customer. If you refine the service, KazanExpress will also be able to use it for:
Search for duplicates in the computer database of goods.
For internal pricing (if many sellers have the product and the price varies greatly).
Search for lost items in stock (by talon, item number).
Similar projects
Chat-bot with computer vision for wallpaper selection at Leroy Merlin
Service with computer vision for plant weight estimation
11/
The system of personal recommendations on the KazanExpress website
