Face recognition system for retail
Summary
As demand grows, retail shops struggle to create the best customer service, since seldom do they completely understand a regular person's needs. This is why our solution provides:
Insight into the customer opinion of the provided service through emotion recognition.
Security for customers and workers through criminal detection based on a provided database of dangerous individuals with the ability to manually update it.
Comprehensive customer analysis including gender and age.
FaceID-like system for keeping track of regular customers.
Project duration:
7 months
Team
5 people
3 deep learning engineers, 1 designer, 1 frontend developer and 1 backend developer.
Technologies
Python, OpenCV, Pytorch, FaceNet, aiohttp, React, nextjs
Tech challenge
Real-time performance with low computational power.
Sustainable data storage with fast I/O.
Easy-to-deploy solution.
Solution
Our solution applies fine-tuned and optimized FaceNet for recognition and customer analysis. For the criminal detection, we use embeddings creation with similarity indexes.
