Why I Chose C++ / Python and Computer Vision

Exploring Machine Learning Beyond the Buzzwords

For basic computer vision tasks — object detection, recognition, segmentation — you don’t actually need to be a programmer anymore. Tools like YOLO already exist. Datasets are ready. Pipelines are built.

But teaching a machine something truly new is a completely different challenge.

I’m not talking about training on COCO or clicking through a GUI. I’m talking about understanding how a neural network learns — and why it sometimes fails. Dimensions, camera angles, lighting conditions, background noise — all of these factors directly affect learning. Knowing what helps a model learn is just as important as knowing what blocks it.

That’s where real engineering begins.

Logic Is More Exciting Than Chess

What truly fascinates me is building logic.

For me, it’s more engaging than playing chess. You explore countless scenarios, connect conditions, optimise paths — and slowly, a system begins to make sense. Sometimes it’s so absorbing that you forget to sleep. And when progress stalls, you know it’s time to step away, take a walk, and let your brain reset.

The best moment is always the same: when the system finally works — and you know you’ve built something meaningful.

Programming languages are not logic.Today, a neural network can write code for you. But it cannot build logic for you.

Working systems require structured thinking — sequences, dependencies, data flow. When different pieces of hardware come together to perform a single task, they must communicate in a precise and predictable way. That coherence doesn’t happen by accident.

My First Project — Learning by Doing

My first real project was built completely from scratch. It was difficult. Frustrating at times. But I was so absorbed that I completed it in just three weeks — using only C++.

That project taught me more than any tutorial ever could. It showed me how deep the gap is between “code that runs” and a system that works.

Where Logic Meets Digital Intelligence

Today, I work at the intersection of two powerful domains:

• Classical automation logic

• Digital intelligence (machine learning & computer vision)

They are very different worlds — but when combined, they unlock entirely new possibilities. Systems that don’t just execute instructions, but perceive, adapt, and respond to real-world conditions.

Looking Ahead

The future of automation lies in this integration.

From healthcare — where machine learning supports diagnosis through medical imaging — to transportation — where computer vision enables safer, smarter vehicles — the applications are vast.

And we are only at the beginning.

Final Thoughts

Choosing C++ and Python for computer vision wasn’t about languages. It was about mindset.

About understanding systems deeply. About building logic that survives the real world. About embracing complexity instead of hiding from it.

And that challenge — is what keeps this work endlessly exciting.

Our digital department: Digital Automations

Thomas J.