Facial Emotion Case Study

Client Overview

In the current fast developing Machine Learning climate, Computers can be very accurately trained to see and understand human sentiment and moods. Optisol has provided human-based Machine Learning to our customers in various fields such as Industrial, Healthcare, Law etc. These solutions detect human activities, gestures, emotions and sentiment.
To highlight one such exemplar, we developed a robust subscription-based web application that recognizes visitor’s faces, age, gender and detects the mood of a person that walks into any retail store. The customer wanted to improve the in store closure rate from 32% and had identified that by being able to read potential customers’ moods when they were in the store would allow for greater success and therefore increased revenue. The solution starts with cameras at the point of capture (in-store). This video feed is streamed to Kinesis Video Stream, then subsequently processed in Microservices Architecture.

Our Approach

null

Configure IP cameras and feed video data to AWS cloud.

null

Stream the video via AWS Kinesis.

null

Use Recognition to detect faces to inform the next step (no faces, no action).

null

Lambda functions orchestrate multiple Recognition services to determine the visitor’s gender, emotions, and age.

null

Lambda functions store the derived data to the RDS

null

The derived data is sent to the appropriate stores as messages.

null

The in-store application displayed the streaming video with the derived metadata for the store staff to see.

The Digital Driver

The implementation of machine learning on the live-streamed video took a total of 3 months to complete and provided the store service staff with great insight to each potential shopper. In combination with appropriate training, the store staff are able to approach shoppers with specific narratives based on their estimated age and emotional states, this resulted in the closure rate rising to 59%.
With the assistance of insights generated, the store owner was able to gauge the overall sentiment of customers and was then able to incrementally bring changes to the store. They measured the effectiveness of the changes (on new overall sentiment). This process of incremental change helped them transform their store with the help of real-time insights from shoppers in a very effective way. Optisol is using the latest advances and cutting-edge technologies in Vision Intelligence to help other clients in this same way.

Architecture Diagram

No-Obligation Consultation
Aws