`Flooring | AI Search Engine | Sicara

BUSINESS CASE

AI-based Search Engine

Challenge

Tarkett, one of the leaders in the floor and wall covering market, wanted to increase its market share thanks to its e-commerce website. However, as the website did not facilitate the buying process for new visitors, we developed a search engine based on image recognition that enhances user experience.

Tarkett is a multinational company, specialised in floor and wall coverings for the health, housing, teaching, commerce, office buildings and sports installations industries, etc. Its clients are architects and installers who rely on the site to place orders and to find new products ideas.

2836M

Revenues in €

13,000

People

92

Localisations

Our impact

12 products
suggested for 1 image uploaded
87%
architects recommend our search engine
Background
Background
Quotes

Sicara played an important role in our differentiation strategy thanks to innovation. Thanks to the team's speed of development and the agile methodology it relied on, we developed an innovative tool in less than 10 weeks, with a real awareness of our customers' needs.

Jean-Hubert Guillot, IT Director – Data / Architecture & Performance @Tarkett

What we did

We implemented a search engine based on image recognition

Maturity Audit, image recognition, AI, artificial intelligence

What we did

We implemented a search engine based on image recognition

We launched this search engine in answer to customers’ needs : in August 2017, 50% searches on Tarkett's website were undertaken with the product reference and resulted in visitors not finding their desired products. Thanks to the image search engine, after uploading a photo of the desired product, the visitor was then offered a selection of similar products. This enhanced the visitor's user experience and Tarkett's conversion rate.

How we did it

With little training data, we relied on data augmentation

How we did it

With little training data, we relied on data augmentation

With little quantity of data to train our dataset, we relied on a strategy of data augmentation as well as photoshoot campaign with our client. We also decided to use Sagemaker, AWS fully managed machine learning service, that was released 5 months earlier, in order to train our neural network on a weekly basis. It was designed to recommend products from Tarkett's 500-product catalogue.

keras, logo, manomano, sicara
logo, python, manomano, sicara
AI, machine learning, deep learning, data augmentation

Our team set up

We worked full speed thanks to agile methodology

startup, sicara, team, teamwork

Our team set up

We worked full speed thanks to agile methodology

Our team was composed of 2 data scientists and 1 agile coach and brought added value to the client on a daily basis.

Geoffroy

HEC

Adil

Centrale Paris

Alexandre

Centrale Paris


Related articles written by Sicara data scientists

blurry street

GAN with Keras: Application to Image Deblurring

A Generative Adversarial Networks tutorial applied to Image Deblurring with the Keras library.

graph

Keras Tutorial: Content Based Image Retrieval Using a Denoising Autoencoder

How to find similar images thanks to Convolutional Denoising Autoencoder.

Edge detection, tutorial, knowledge

Edge Detection in Opencv 4.0, A 15 Minutes Tutorial

This tutorial will teach you, with examples, two OpenCV techniques in python to deal with edge detection.