Junaid Iqbal | Agentic AI Engineer

In the heart of industrial hubs like SITE and Korangi, the steady rhythm of the power loom is the heartbeat of our economy. But as any textile engineer knows, that rhythm can be interrupted in a split second by a single broken thread or a tiny hole.

For decades, we’ve relied on the human eye to catch these flaws. But eyes get tired, shifts are long, and high-speed production wait for no one. That is why I decided to build the Loom-Guardian—a simple, smart, and hardware-friendly AI system designed to watch the fabric when we can’t.

Why “Logic-First” Engineering Matters

Most AI projects today require expensive computers and massive power. But in a real factory, we need solutions that work on the tools we already have. I developed this project on my Dell Latitude 5490, focusing on Resource Efficiency.

Instead of just “throwing AI” at the problem, I used a Logic-First method:

1. The Textile Eye: Using Adaptive Vision to handle the flickering lights and shadows of a real factory floor.
2. The 5-Frame Rule: A “Temporal Buffer” that prevents the system from crying wolf over a bit of dust or lint. It only alerts the manager if the defect is real and persistent.
3. The Audit Trail: Every time a hole is found, the system logs the exact second and saves a photo. This isn’t just a monitor; it’s a digital record for quality assurance.

Keeping it Simple and Secure

I wanted this project to be accessible. Whether you are a student, a factory manager, or a fellow developer, the method is easy to follow:

• No Heavy Hardware: It runs purely on a standard CPU.
• Privacy Minded: All processing happens locally on the machine.
• Transparent: Every line of code is written to be readable and maintainable.

Explore the Project

I believe in open-source learning and collaboration. I have prepared all the documentation you need to understand or replicate this work:

• [Full README File]: For a deep dive into the technical setup and environment.
• [Project Presentation (PDF)]: A slide-by-slide breakdown of the industrial impact and ROI.
• [GitHub Repository]: Review the code, suggest improvements, or star the project for your own use.

---

Let’s Connect

Engineering is about solving problems together. Whether you are looking to implement a similar Computer Vision solution in your facility, or you want to discuss the intersection of Textile Engineering and Data Science, I am always open to a conversation.

Are you working on an industrial AI challenge? I am currently available for consultations and project discussions. Let’s build something that makes our industries smarter and more efficient.

Contact Me via LinkedIn

and email: junaid19tex@gmail.com