India’s massive road network, the second-largest globally, has traditionally relied on manual tollbooths, leading to congestion and long commute times. To address this issue, Calsoft, an Indian- American tech company, implemented a solution powered by Nvidia technologies. The system integrates with India’s dominant payment system, UPI, to automate toll collection.
A key challenge was the diverse range of license plates in India, which posed difficulties for automatic number plate recognition (ANPR) systems. To overcome this, Calsoft built a solution that automatically reads passing vehicle plates and charges drivers’ UPI accounts. This approach reduces manual toll collection and helps address traffic congestion.
The solution has been deployed in several major cities as part of a pilot program, achieving an accuracy rate of around 95% through the use of an ANPR pipeline. Nvidia’s technology played a crucial role, particularly in improving night-time detection and model accuracy despite environmental factors like fog, heavy rains, and reflections from bright sunshine.
Calsoft used Nvidia Metropolis to track and detect vehicles throughout the process. The team also deployed Nvidia Triton to manage AI models and built a real-time streaming platform using Nvidia DeepStream software development kit. This allowed for efficient processing and analysis of data streams, incorporating advanced capabilities like real-time object detection and classification.
The solution uses Nvidia hardware, including Jetson edge AI modules and A100 Tensor Core GPUs. It is also scalable, designed to accommodate future growth and expansion needs, ensuring sustained performance and adaptability as traffic conditions evolve.
Source: https://venturebeat.com/games/how-india-uses-nvidia-accelerated-computing-to-ease-tollbooth-traffic/