Real Time Facial Analysis For Digital Signage

Problem

Our client faced the challenge of accurately detecting and analyzing facial features in real-time. They needed a solution that could handle many detected faces in different lighting conditions and provide robust facial features extraction system.Features that should be provided to centralized streaming server in near real time where age, gender, and emotions.

Solution

Our team addressed the problem by integrating custom image-processing application, which utilize deep learning frameworks. Integrated application collects frames from the camera and sends them for further processing to Age/Gender Service which was integrated as a optimized neural network wrapped in a gRPC service, which predicts the age, gender and emotions of the persons in the captured frames. Results of real time analytics is used by streaming service to adjust content played on digital signage displays. Batch analytics is used to provide statistical data and tailor the digital content which is presented.

Key Metrics/Technologies

The outcomes of the solution are highly accurate facial analysis in near real time. Age, gender, and expressions detection provided accurate results, enabling automatic determination of gender and age group in front of specific display without the need for manual input.

Some of the technologies used on this project include:

• Nvidia Jetson nano
• OpenCV
• Google Cloud
• Keras
• Pytorch

Client

The client is a Swiss manufacturer of displays and streaming solutions for digital signage.