Resliv
Machine learning solution development using Python Keras framework
FashionTech
The building of an ML platform for FashionTech startup
Our project’s highlights
- Development and training of AI engine
- Development of the Business Rules Engine
- Development and integration API and widgets
- Mobile SDK for iOS and Android
The Client
Since 2016 a Singapore SuitApp startup provides its services in the field of FashionTech.
The client approached Resliv to develop a machine learning platform that would provide a smart, personalized recommendation service automatically, creating outfits from the goods of any fashion-online store.
The Challenge
To increase the engagement rate and broaden the circle of customers, the client needed the app that could precisely recognise objects on images and create outfits. Each chosen item supposed to have 3-5 looks for different styles and occasions.
The solution also supposed to constantly auto-update prices, sizes and availability of the goods.
Solution Overview
Getting down to work, our dedicated team of software engineers developed a back-end, based on Machine Learning algorithms.
- With the objective to train the system to recognise items on the images, the solution should also determine 30+ the parameters (colour, print, the length of a sleeve, etc.) and classify objects according to them.
- Next step was to learn the system to generate 3-5 outfits with a chosen item.
- Assuming that fashion trends update seasonally, the software should contain a major database and a set of business rules that would determine perfect matches.
- As the solution supposed to easily integrate with any online shop, Resliv also developed API, widgets, and mobile SDK for iOS and Android. Thus, the product can be used within the product page, as a separate page, in a virtual mirror app or for personalized based targeting in social media.
Results
– Generation of an unlimited number of outfits
– Increased Conversion Rate, because of an extremely user-friendly 24/7 service that increases click on recommendation +40%, conversion by 30%
– Post-sale and Up-sale conversion increase as part of email marketing and average purchase size increase
– 3X increase of product view per session, which, in its turn, leads to higher engagement
– Increase of CRR (Customer Retention Rate) – 35% higher repeat visit rate.
Scope of Services
Analysis
Development
Testing and QA
Deployment
Support
Technologies
Python, Flask
Tensorflow, Keras