Real time car make and model detection for Indian market

Automatic cars make and model detection and tracking play a significant role in the areas of traffic control and management, law and order, and market research. With advances in camera technology and embedded chips, Computer vision a subset of AI is heavily used for safety and security use cases. This work examines how the YOLOv4-tiny computer vision algorithm can be used to identify and locate Indian car brands and models from real-time video streams. The model has been trained with the latest YOLOv4-tiny algorithm for 21 Indian car brands with 23k images. Input for model prediction will be a 30-sec user recorded video with multiple car models in 25 frames per second, and the output generated is video content with annotations to their make, model, and bounding boxes. Apart from safety and security, the model can be enhanced further for different commercial brands and models. This would unlock enormous business opportunities in the market research and marketing field (e.g.: Shoppable content) to create an ultimate user experience.