Skip to content

AI-powered industrial vision system summary using CLIP (ViT-B/32) and Modbus TCP for PLC integration. (Overview - No Code)

Notifications You must be signed in to change notification settings

misolmaz/VisualIQ-Industrial-Vision-System-Overview

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 

Repository files navigation

🌟 Highlight Project: VisualIQ – Industrial Product Recognition System

💡 Overview

VisualIQ is an AI-powered industrial vision system designed to automate product recognition and verification processes on production lines. It integrates high-speed camera acquisition, advanced AI analysis, and PLC communication for reliable, human-intervention-free production management.

Core Problem Solved

Manual or barcode-based product verification is slow and error-prone. VisualIQ provides a high-speed, flexible solution for product verification, labeling, and real-time production reporting across diverse manufacturing environments.

⚙️ Technical Architecture

This system demonstrates expertise in full-stack industrial integration, bridging Python AI models with industrial hardware. VisualIQ System Architecture Diagram

1. AI & Vision Pipeline

  • Model: OpenAI CLIP (Contrastive Language–Image Pretraining) ViT-B/32 is used for robust visual recognition.
  • Self-Learning (Incremental Training): New product images are collected automatically (dataset/train/{ürün_kodu}). Once sufficient data is gathered, the system notifies the operator for approval, triggering incremental training and integration into the clip_vector_db.npz vector database.
  • Data Scope: Operates on an active product list retrieved from the vw_SonUretimEmri view in a SQL Server database, ensuring recognition is limited to currently relevant items.
  • Camera Interface: Utilizes an optimized DMV SDK interface for high-speed industrial camera acquisition, with a dedicated simulation mode available.

2. Industrial Integration

  • PLC Communication: Receives a trigger signal from the PLC and transmits the identified product code back using the Modbus TCP protocol.
  • Software Stack: Developed entirely on Python + PyTorch.

3. Operator Interface & Management

The Human-Machine Interface (HMI) is built with CustomTkinter for a robust desktop experience, featuring:

  • Training Mode: Manages automated new product image collection and approval-based model updates.
  • Active Products & Log Viewing: Provides real-time tracking, error logs (JSONL format), and production statistics.
  • Management Panel: Centralized control for system configuration.

✨ Key Technical Achievements

VisualIQ provides a high-speed, accurate, and sustainable industrial AI solution by seamlessly integrating Camera – PLC – Database – Artificial Intelligence into a single, manageable platform.


⚠️ IMPORTANT NOTE ON CODE ACCESS

This project is a high-value commercial application. Due to proprietary agreements and commercial secrecy, the source code is kept in a private repository.

If you are a serious hiring manager or technical lead, please feel free to contact me via https://www.linkedin.com/in/misolmaz/ for a confidential technical discussion or a demonstration of the system's architecture and performance.

About

AI-powered industrial vision system summary using CLIP (ViT-B/32) and Modbus TCP for PLC integration. (Overview - No Code)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published