CHALLENGE:To be honest, there was no problem as such. But today the ocean of competition in services is no longer even red but already crimson, so one of the main tasks of any business is to improve its services in all available ways.That’s why we started cooperating with one bank, whose name is under NDA, which wanted to evaluate its services using Computer Vision. We have to tell that this project in the Security Systems field has become one of the most fascinating and interesting that we’ve ever participated in.TASK:The goal was to teach cash dispensers to recognize people’s emotions when using the device assessing the quality of the services provided. This was possible thanks to Computer Vision technologies, so we began to generate ideas and select the most viable. See what we got. SOLUTION: Using Deep Learning, Computer Vision, Python, and TensorFlow we created a model that can recognize 7 basic emotions of a person like joy, surprise, sadness, anger, disgust, contempt, and fear.This is how it works. We integrate our soft in ATMs systems. Each device has a camera which marks and analyzes facial expressions through the movement of parts of the face. According to this data, our model determines the emotion. On this base, the ATM knows how its user feels and send this information to the data center where it will be processed and converted to conclusions and statistics.RESULT:The model we developed recognizes emotions with 97% accuracy according to our tests. This enables our customer to estimate whether their services make the users feel pleased.Our soft had just been integrated so we do not have any facts on how the bank improved its services but we will immediately update as soon as we receive this data.
