Well functioning traffic flows, real-time parking availability, increased security and swift crime solving, cost-efficient defect detection in the infrastructure or updates on road conditions – smart cities could already be taking advantage of real-time fully automated video surveillance from drones, ground vehicles with video cameras and CCTV, Lithuanian startup PIXEVIA claims.
Due to the company’s innovative methods in Artificial Neural Network learning, its AI based real-time computer vision system enables autonomous data processing and information fusion, providing drone manufacturers, software developers and smart cities with tools to increase efficiency of their services or life quality for their citizens.
PIXEVIA is able to identify different objects like cars, their license plates, size, speed or coordinates, people, and infrastructure objects from multiple video footage in real-time. Application of this AI system is boundless: it can be adopted for surveillance and security purposes, powerline-pipeline inspections, inventory management and transportation. But most importantly, the system is highly user-friendly and adaptable. It’s made up of step-by-step functional components, which can be adjusted to user’s needs without requiring specific programming or AI knowledge from the user”, says Mindaugas Eglinskas, PIXEVIA creator.
Due to PIXEVIA’s elegant architecture, the system can be extended with new modules and neural networks to work with various tasks. The system is already equipped with a wide range of primary technological functionalities but it can be further tailored to the user’s needs. The system requires only the minimum level of human supervision when in use, as all the components of the system pursue their tasks autonomously.
The other source of system’s adaptability lies within PIXEVIA’s deep neural networks, which can be specially trained using computer simulation of, for example, different car or ship models, in result reducing manual human work in data labeling. The system is also self-adapting after the deployment: the information gathered by the fast neural network can be compared with results from a large, accurate and slow machine learning or computer vision algorithms during the “sleep” period, making it possible for fast networks increase the recognition quality or recognize new objects.
AI based computer vision systems provide a unique opportunity for municipalities and other city services to benefit their citizens. For example, it can analyze cars in parking lots, notifying drivers on empty parking spaces and parking alternatives; predict best places for new shops by analyzing traffic flows; or assist in solving crimes faster by quickly recognizing and automatically tracking suspected individuals on an extended amount of video footage, increasing safety and life quality in the cities”, says Mr Eglinskas.
Sign up for the free insideAI News newsletter.
