Machine Vision - Python/Roboflow

This project implements a real-time machine vision system to detect and count U.S. coins using an Android smartphone camera and a custom Roboflow-trained model.

Overview

  • Trained a Roboflow model to detect and classify U.S. coins.
  • Streamed images from an Android camera to a Python script running on a PC.
  • Detected and labeled coins in real time, calculating the total value and overlaying bounding boxes.
  • Captured snapshots or video of the live output for analysis.

Demonstration Video

The following video shows multiple test runs using the Python script, demonstrating real-time detection, labeling, and coin value calculation:

Try the Model (Prototype)

You can experiment with the Roboflow-trained model directly from your device using this QR code:

QR Code

Note: This live model is a prototype. For best results, use consistent lighting and maintain a moderate camera distance. The full Python script is included in the repository for reproducibility.

Project Files

The complete source code is available on GitHub:

→ usd-coin-counter Repository


← Back to University Projects