FactoryVision is an AI-powered object detection platform built for conveyor-belt factories to automate inspection, counting, and defect detection in fast-moving production environments. Using YOLOv26 and OpenCV, the system analyzes camera feeds to identify products, verify condition, and record traceable results in a local database. Designed with a fault-tolerant CCU-integrated architecture, it operates fully offline without internet and buffers events for later cloud synchronization. Built to reduce manual inspection, prevent quality escapes, and maintain reliable production visibility through resilient edge-to-cloud processing at industrial scale.

Tech Stack

• YOLOv26
• OpenCV
• Node.js
• Edge-to-Cloud Architecture

Key Features

• Real-time object detection, counting, and defect analysis on conveyor belts.
• Offline-capable architecture with local buffering and event persistence.
• Automatic cloud synchronization with retry logic after connectivity restoration.
• Traceable inspection records with alerting for defective product events.

My Role

I was responsible for designing and developing the edge-to-cloud detection pipeline, implementing offline local queuing and sync retries, integrating AI vision processing, and building reliable data flows for zero-loss industrial traceability.

Challenges & Solutions

The main challenge was building a fault-tolerant vision pipeline that could operate fully offline without losing inspection data. Coordinating real-time detection, local database writes, buffered event queues, cloud sync retries, and production alerts required careful architecture to ensure traceability and uninterrupted factory operations.