Object Detection REST API for Retail Shelf Imagery

Retail operations teams increasingly automate shelf audits, planogram checks, and stock monitoring, and all of those workflows begin with the same primitive: given a photo from a store, what objects are in it and where? Rather than rebuilding detection logic inside every application, retailers need it packaged once as a clean, reusable service any internal tool can call. The intern builds exactly that service with FastAPI: a REST API that accepts image uploads from retail environments, runs a pre-trained YOLO model in PyTorch, and returns structured JSON describing each detected object with its bounding box coordinates, class label, and confidence score. OpenCV handles image decoding and preprocessing on the way into the model. The intern designs the API contract carefully, covering request validation, error responses, and interactive endpoint documentation, and adds Docker packaging so the detector deploys identically on a laptop or a cloud host. This project teaches the machine-learning-as-a-service pattern that dominates production AI: separating model inference from user interfaces behind a stable API boundary. The intern demonstrates skills in modern Python API development, deep learning inference integration, and shipping a component other engineers could adopt immediately.