Calsoft and NVIDIA Bring Vision AI to Solve India’s Tollbooth Traffic

In partnership with NVIDIA, Indian-American tech company Calsoft is automating tollbooths across India using advanced vision AI. With the world's second-largest road network spanning nearly 4 million miles and over 1,000 tollbooths, many of which operate manually and contribute to traffic delays and congestion, this new solution aims to significantly ease the problem.

Automating Tollbooths for Efficiency

Calsoft's innovative solution integrates NVIDIA's technologies with India’s Unified Payments Interface (UPI) system to streamline toll collection. One of the primary challenges in automating India's tollbooths is the country's non-standardised vehicle license plates, which vary in color, size, font, and language. These variations posed a significant obstacle to the accuracy of Automatic Number Plate Recognition (ANPR) systems.

Despite these challenges, Calsoft has developed a system that reads vehicle plates automatically and charges the associated UPI account, reducing the need for manual toll collection. This approach represents a major advancement in managing highway traffic.

AI for License Plate Detection

The solution, currently being piloted in several major Indian cities, boasts a 95% accuracy rate in reading plates. This achievement is largely due to the ANPR pipeline, which detects and classifies license plates as vehicles pass through tollbooths. NVIDIA's technology has played a critical role in overcoming obstacles such as night-time detection and environmental impacts like fog, heavy rain, and bright sunlight, which can distort plate recognition.

Vipin Shankar, Calsoft’s Senior Vice President of Technology, emphasised the challenges faced and how NVIDIA Metropolis, an AI application framework, was essential in enhancing the system's operational efficiency. 

““Particularly challenging was night-time detection. Another challenge was model accuracy improvement on pixel distortions due to environmental impacts like fog, heavy rains, reflections due to bright sunshine, dusty winds and more,” he said. 

Calsoft also employed NVIDIA Triton to deploy and manage their AI models and utilised the NVIDIA DeepStream SDK to build a real-time streaming platform capable of advanced object detection and classification.

Scalable Solutions for the Future

Calsoft's tollbooth automation solution, powered by NVIDIA Jetson edge AI modules and NVIDIA A100 Tensor Core GPUs, is designed to be scalable, ensuring it can accommodate future growth and evolving traffic conditions. This development marks a significant step forward in India's efforts to ease road congestion and improve the efficiency of its highway network.

Interestingly, Union Minister Nitin Gadkari recently said that the country will move towards a new form of satellite-based toll collection system, where money will be charged from one’s bank account and the amount of road covered will be charged accordingly. This step will help towards saving time and money.