The neural network analyzes the existing database of unwanted persons’ photographs and compares them with images from online cameras. If the images from the camera and the database match, an instant notification is sent to the mobile device that a person has been detected, indicating the camera that recorded that person for quick security response.
The system can also analyze faces and, if a person can be dangerous to the organization (by physiognomic characteristics), issues a notification to the security service.
- Develop a system using machine learning without using ready-made learning algorithms.
A new system was developed for the project, which fully fulfilled requests trained to an accuracy of 80-94% of face recognition. Accuracy lies in the quality of a camera that transmits an image to artificial intelligence.
The scientific data on facial features and their correlation with the commission of offenses were also analyzed and implemented into the system.
While working with the prototype, the system analyzed 200 people and issued 2 alarms. One of which was a man who was involved in hooliganism in his youth. The system correctly analyzes faces from the crowd and provides security guards with the ability to respond to potential threats.
- 3 Python developers
- Machine learning expert (Ph.D)
- Ph.D Psychologist
- 2 Ionic developers
- Angular developer
- QA engineer