What it is
Source: My own description of the project; IBM and Wikipedia for the definitions. Date: 15 October 2026. Scope: What the app does, how it is built and where it runs. No model, accuracy figure or frame rate is claimed, and no factory or client is named.
The Industrial Vision App analyses factory video in real time, as it plays, and runs locally on Windows. It detects and tracks objects, flags anomalies and shows a live dashboard. Built
What it does
- Real-time analysis. The video is analysed while it plays, not after it has been processed in full.
- Detection and tracking. It detects objects, locating and classifying them (IBM), and gives each one a persistent tracking ID, so the same object keeps one identity from frame to frame (Wikipedia).
- Anomaly flags. It flags visual and behavioural anomalies, events that deviate from what is usual, standard or expected (IBM).
- Live dashboard. A dashboard of genuine metrics that updates live while the video is analysed.
How it is built
It is written in Python, and inference runs locally on Windows, so the models run on the machine itself rather than on a remote server. Built Its models and thresholds are configurable, and the video source can be swapped for live cameras.
Stack. Python, local inference, Windows
Where it stands
It does what is described above, on a local Windows machine. Built No accuracy figure or frame rate is claimed, and no factory or client is named.
My part
I designed and built it. Built
