• This AI-driven Node.js based car model recognition system recognises car model using UNet and ResNet-50 Deep Neural Network models.
• Models are trained using IPython kernel for fast training speed.
• Used 110000+ images for UNET training (accuracy 98+%).
• For ResNet-50 training we will use web crawler.
• System's architecture is unique, it has 3 parts: mask generation, boundary box cropping and classifier.