About Customer
Customer is a big Australian Company that specializes on the variety of retail services for Retail Industry. The Customer has developed groundbreaking services for numerous product categories, including but not limited to general merchandise, apparel, household appliances, groceries, and consumer electronics. The Customer offers customized services to cater to the specific requirements of individual clients, and its solutions are utilized by a wide range of customers worldwide, ranging from small store owners to large multinational retailers, with a global trading footprint.
Challenge
The Customer made the strategic decision to implement a cutting-edge solution that would leverage advanced facial recognition technology to enable real-time data capture and analysis of customer metrics. The project aimed to seamlessly integrate with the existing IT infrastructure, allowing for the creation of a centralized database that would store and manage the collected data. By utilizing machine learning algorithms, the system would be able to identify returning customers based on their facial features, facilitating enhanced personalization and delivering a seamless omnichannel experience.
Solution
To deliver a high-quality solution for the Customer, Spybrick’s team of experts applied their expertise in various areas of computer vision, including image processing, deep learning, and data analysis.
To enable face image capturing, we employed OpenCV, an open-source computer vision and machine learning software library, as a foundation platform. Our team optimized the library’s image capturing capabilities to provide the best possible results for facial recognition.
In the second stage of development, our team implemented a complex process to enable effective facial recognition. Firstly, we utilized preprocessed input images, which had been specifically processed to make customers’ faces recognizable, as the basis for facial recognition. Secondly, we calculated landmarks, i.e., specific facial features, using advanced algorithms and techniques such as Harris corner detector, Flusser affine invariants, and Hu invariants. Finally, we utilized a set of sophisticated algorithms, such as Mahalanobis metric and SIFT (Scale-invariant feature transform), to facilitate face recognition based on the calculated landmarks and special criteria.
Throughout the project, our team extensively researched various algorithms and methods to ensure the highest level of accuracy and efficiency. The combination of these advanced techniques and methodologies allowed us to achieve the Customer’s requirements and develop a highly effective system with a significant percentage of correct results.
Results
Thanks to our efficient project management and technical expertise, Spybrick delivered the solution to the Customer on time and within budget. The solution was designed to enable the use of innovative image recognition technology, allowing the Customer to recognize and identify store visitors with precision. With this information, the system could tailor each customer’s in-store experience to their preferences and enhance their overall shopping experience. The solution utilized state-of-the-art machine learning and computer vision algorithms, ensuring optimal performance and accuracy in recognizing store visitors.
Technologies and Tools
.Net 2.0/3.5, C++, IPP, OpenCV library.
