Project 04 · Computer Vision · Simulation

AI-Powered Industrial Vision

Real-time machine vision simulation replicating an industrial conveyor belt inspection station with automated defect detection and rejection.

View on GitHub ↗

Overview

AI-Powered Industrial Vision Inspection System is a real-time machine vision simulation that replicates an industrial conveyor belt inspection station. It combines Unity for digital twin simulation, FastAPI for backend communication, and OpenCV for computer vision-based classification.

The system automatically captures frames of packages moving along a simulated conveyor belt, processes them through a computer vision pipeline to determine PASS or FAIL status, and triggers a simulated reject mechanism for defective packages — demonstrating a complete industry-grade inspection workflow.

Architecture

Unity Digital Twin (Conveyor Simulation) ↓ Frame Capture & REST API Transmission ↓ FastAPI Backend Server ↓ OpenCV Image Processing Pipeline ↓ PASS / FAIL Classification ↓ Response to Unity Client ↓ Automated Reject Mechanism

System Demonstrations

Full demonstration of the industrial vision inspection pipeline — package detection, classification, and automated rejection.

Extended demo showing the conveyor belt simulation with multiple package types and real-time PASS/FAIL classification.

Key Features

Digital Twin Simulation: Unity-based conveyor belt environment that accurately replicates industrial inspection conditions.
Real-time Frame Capture: Automated screenshot capture of packages as they pass the inspection station.
CV-Based Classification: OpenCV image processing pipeline for analyzing package quality and defect detection.
REST API Communication: FastAPI server bridging the Unity simulation and Python CV backend with low-latency responses.
Automated Rejection: Simulated reject mechanism triggered automatically for packages classified as FAIL.
Production Workflow: End-to-end pipeline demonstrating a realistic industrial quality control workflow.

Tech Stack

Frontend & Digital Twin

Unity 3D C# Inspection HUD

Backend

Python FastAPI REST API

Database & Storage

SQLite Inspection Logs

AI & Computer Vision

OpenCV Machine Vision Defect Detection