Retail & eCommerce
Python, Google Cloud, .NET
The client is a large photography agency that started as a single photographer and has since grown to a business that provides services across the country. They came to NIX to create a system that would improve their teamwork and navigate an enormous photo library.
It would significantly reduce time spent on routine tasks and automate a number of processes.
The client’s photo library, increasing in size over many years, has become a massive database of photos with complicated internal navigation. A couple of times before, the client tried to implement a categorization system, but these attempts failed. We needed to develop a solution that will decrease the time that photographers spend on browsing folders to find necessary images, which in turn included the following goals:
Develop an entire web platform that will serve as an interface for managing a cloud computing system based on Google Cloud Platform
Create and train AI computer vision algorithms for face and emotion recognition to simplify image cataloguing and decrease time on routine
Using .NET as a central technology, we developed a web platform with a simple and intuitive interface based on Google Cloud Platform.
The platform provided functionality for two types of users: photographers and their clients. By default, both of them can use the main features, namely face and emotion recognition. However, they have different access to data management. A photographer’s account allows users to upload pictures, create a data pool in which the system searches for similarities, and recognizes expressions. Users can create several projects in the account and share access to their content with a unique link for convenience.
Moreover, if users noticed that the computing system made a mistake in face or emotion recognition, they can manually point the system to correct results; thus, the algorithm can evolve continuously.
With a client’s account, users can only view the project’s content using links from the photographer and use recognition features.
First, users pick an instance by selecting a photo in the cloud database or uploading it.
After that, the system, through an API, launches a recognizing process and scans other images in the database to find people from the selected instance. It’s worth mentioning that users can choose photos with several people, and the system finds all images with at least one of those people.
The platform recognizes the emotion people express in the photos, showing emoji and percent results. The developed system is capable of recognizing up to eight different emotions.
Users also can search all photos in which people express the same emotion, such as finding all images with smiling people.
The first task NIX engineers had to solve was to consider what face detection technology to use.
Based on our extensive experience in building image recognition models, we decided to use RetinaFace, one of the most advanced face detectors worldwide to achieve better accuracy compared to the traditional convolutional neural network (CNN).
There was no need to train a recognition model from scratch and spend additional effort in vain. Instead of this, we used a ready-made model, already trained on the VGGFace2 dataset, which contains 3.31 million images of 9131 subjects (identities). Data engineers simply configured the model’s topology to cover project needs optimally. After that, the model training continued, using AffectNet, a facial expression database.
As a final touch, we polished the model using high-quality photos with strong emotional expressions of more than 2000 NIX experts to tune facial expression recognition.
The smart web platform significantly accelerates the client team’s routine tasks, allowing them to spend more time shooting instead of browsing folders. Currently, photographers rate this innovation highly.
Furthermore, our client is considering updating FaceMe and making it into a commercial product as a service.
7 experts (Project Manager, Business Analyst, Backend Developer, Data Science Developer, Markup Developer, QA Engineer, Designer)
C#, Python, TensorFlow, Keras, Google Cloud, Kubernetes, React.js, TypeScript, Firebase, .NET, SCSS
NIX helped the client to inform prospects about the new business line and got a 460% revenue growth and a 354% increase of ROAS within six months.
Email marketing activities increased customer retention and grew email marketing revenue by three times.
NIX team elevated the customer shopping journey with intuitive navigation, visually appealing interfaces, and seamless functionality.
SEO activities to enhance website structure and attract more organic users necessary for the company’s consistent growth.
Attracting users to the app and encouraging them to make regular in-app purchases resulting in 100k+ downloads within four months.
How we developed an ecommerce website to provide US companies with white-label, custom-packaged coffee.
Implementing business automation solutions and boosting the entire digital transformation of the nationwide retail chain.
eCommerce platform for improving logistics processes and retailer’s experience through tech-driven services.
Credit-as-a-Service solution connected brands, merchants, and buyers and provided them with unique shopping & selling experience.
See more success stories
Our representative gets in touch with you within 24 hours.
We delve into your business needs and our expert team drafts the optimal solution for your project.
You receive a proposal with estimated effort, project timeline and recommended team structure.