
The Challenge: Outdated Methods and High Operational Costs
Norway’s legacy tolling and parking systems were plagued by inefficiencies and revenue losses caused by outdated fee collection methods. These systems, relying on trust-based models or physical barriers, failed to ensure all users paid their dues. This limited the ability to fund essential maintenance for roads and parking areas.
Finter AS, a leading technology company specializing in smart mobility, recognized these challenges in its operations. However, their existing setup, which relied on multiple CPUs and PCs using classic computer vision techniques, presented numerous operational difficulties:
- Excessive heat generation.
- High energy consumption.
- The need for bulky technical cabinets unsuitable for complex AI workloads.
As AI technology advanced, Finter sought to evolve its systems to harness machine learning and computer vision capabilities. However, the limitations of their CPU-based platform rendered advanced AI workloads impractical. Finter needed a solution that could efficiently handle modern AI algorithms while reducing system complexity and operational costs.