VTechCV

Today’s intelligent buyers respond via social media to show their support for certain products.  Being able to capture these sentiments in real time in-store can convert that data into revenue—sale in a speedy fashion.  With this in mind, Villa-Tech has a goal to allow the industry to adopt deep learning and advanced analytics, and with VTechCV, we can use deep learning and advanced analytics to understand the buyer’s sentiment for richer insight, resulting in higher sales and best business practice.

Being able to track customer reactions, responses, and nonverbal cues in retail environments greatly adds depth to binary information.  Thanks to the research of Alex Pentland from MIT’s Human Dynamics Laboratory and Entrepreneurship Program, we have grasp on to an approach called social physics, a theory about understanding how the flow of ideas and information translates into behavioral changes.  These changes can only be caught using ubiquitous digital data call records, credit card transactions, GPS location, and video surveillance.

The goal for our computer vision, VTechCV, is to be a new disruptive solution to process customer sentiments when they are engaging with the retail environment and when they engage with the environment’s products.  With this, we can allow big data to see customer responses while creating richer insight.

VTechCV’s main function is to identify and categorize a customer’s emotional expressions.  Here’s a step-by-step description of how it works:

1. Human Recognition (via live video).  Our platform detects a person in a live video feed using the YOLO algorithm, a benchmark in object detection.  It can detect a person in real-time and up to 95% accuracy.

2. Facial Recognition.  Once the person is detected, their face is detected using the HOG algorithm (Histogram of Oriented Gradients), which is the most popular algorithm for facial detection.  We have trained this platform with millions of facial images collected from Internet.

3. Emotional Recognition. After detecting the face,  that portion of the image is fed to our trained VGG neural network, trained with millions of images to recognize, analyze, and categorize the face’s emotional expression.  There are currently 6 recognizable emotions on our platform: Happy, Sad, Neutral, Angry, Surprised, and Disgust.

With VTechCV, retailers can increase sales by detecting the consumer sentiment and deliver an engaging in-store experience in real time.