Our AI platform makes it easier, more cost-efficient and faster to arrive at the right solution. The platform is versatile and provides a basic structure for implementation of AI models that can be adapted for new applications depending on user needs and system requirements.

The platform allows for several types of customizations and optimisations:

• Customizable feature set such as object detection, classification and anomaly detection
• Flexibility in HW choice for edge computing
• Optimisation of the AI models for the system
• Inclusion of signal processing and other mathematical models for system functionality and optimisation

AI platform development principles

Qamcom AI platform can easily be configured to accommodate a large spectrum of use cases. It was designed to be:

• Configurable:
  • A model zoo of different AI algorithms is implemented in the platform, from which a suitable model can be selected based on requirements such as accuracy and speed.
  • The platform can be adapted to different software architectures depending on the requirements on latency, privacy, cost and security; this includes edge-to-edge and server-to-client architectures.

• Modular:
  • The development of AI algorithms is highly modularized. It supports the selection and removal of different existing modules such as object detection, tracking and counting.
  • Whenever a project requires specific functionalities such as e.g. 5G communication, customized modules can be developed and easily added to the existing ones.

• Portable:
  • The platform can be deployed on different hardware platforms depending on the use case and requirements. For instance, the AI algorithms have been fully customized and deployed on Nvidia Jetson GPU platforms to meet the requirements of limited payload and power consumption.
  • To make the software compatible with different operating systems, the platform supports container-based virtualization.

Solution development balancing performance, cost and energy consumption

Example applications