SnapRapid

Our client asked us to build a complex brand-recognition analytical platform that would help small and worldwide brands track their brand visibility and analyze its performance and advertising payback. As a result of our work, the client received a Data Science solution that analyzes brand name mentions in the text format in social media. Most important, the platform has an advanced video recognition functionality that detects brand logo visibility, a number of times logo was in the frame during the video, etc. The system analyzes the influence of logo appearance on the required video (for example, a live football game broadcast) and connects this data with the number of viewers in certain regions, languages, etc., and arranges all collected and processed data in reports with all kinds of interactive charts.

This platform is made for the B2B sector and implies an individual approach to its monetization. Prior to launching a Big Data mining and processing project for a new client, managers learn the client’s requirements to the smallest details. The client is charged according to the requirements, processes complexity, reports extensiveness, chart types, etc.

To meet the client’s requirements, we combined the most efficient technologies, microservice architecture that makes the system scalable only where it needs to be, and complex cloud infrastructure. We built the platform from scratch, and to make it the most precise and effective, we developed multiple unique algorithms for graphics processing and indices construction. Text and video parsers are written in Python and are based on AI and ML solutions and can recognize whether a brand name or logo is mentioned in the positive, negative, or neutral context. We used Scrapy, Clickhouse, and other technologies to build different types of scrapers for the platform. The client-side dashboard is built with Symfony PHP and AngularJS that are perfect for visualizing processed data as all kinds of interactive and stable charts, including real-time heat maps. The admin panel was created with the combination of Python, Eve REST API, and AngularJS.