The goal of this system is to provide a lightweight inference API that can be deployed on low-resource devices (like Raspberry Pi or Jetson Nano) to classify images captured by embedded cameras. It balances the trade-off between model accuracy and inference latency.