Artificial intelligence (AI) has become increasingly relevant in the field of mobility, particularly when it comes to traffic detection and management. With the rise of smart cities and autonomous vehicles, AI plays a crucial role in ensuring safe and efficient transportation. At NovelSense, we recognize the importance of AI in the mobility of the future and have been working on optimizing our neural network (NN) architecture to improve vehicle tracking and classification. In this blog post, we’ll share more about our efforts and the findings of the SDIL project “NeuralArchSearch.”
When it comes to utilizing artificial intelligence (AI) for traffic detection, one of the biggest factors that affects its performance is hardware restrictions. Our ABAKUS.AI system makes use of small, low-power device that can be retrofitted for smart cities. The device is primarily used for traffic counting and other mobility-related use-cases that require vehicle tracking and classification.
As with all low-power systems, there are challenges related to latency, power consumption, and accuracy in vehicle tracking and classification. To address these issues, we set up the SDIL project “NeuralArchSearch”. The project helped us to test different modes to significantly reduce latency while maintaining accuracy with a deployable architecture modification.
For more details please view the video below: