Spark plug reader

2024

Aim of the project: Create a tool that uses image recognition capable of analyzing the carburetion of a Two-Stroke engine based on its spark plug.

Description: Knowing the carburetion of a vehicle without a video or without trying it is near impossible,

but for before that you can use the spark plug to have a preview of the situation.

A lot of friends of mine with two-stroke bikes was sending me photos of their spark plugs to "read" their carburetions since I wasn't in my hometown.

I developed this tool for them to solve their problem about finding out the right carburetion.

This project originally started with the idea to read the piston crown, however open a kart engine and take a picture of the piston is an easy and fast operation; opening a motorcycle engine isn't that easy and fast.

Also training a model to suit this requirements requires a lot more images for having a working tool, as transfers and exhaust port can change from engine to engine and produce different results.

Using Python with OpenCV, NumPy and other few libraries I was able to create a basic software that is able to recognize the carburetion of a two-stroke engine based on the color of the spark plug.

I've also insert a feature that will recognize whether the image is a spark plug or not.

Results: A simple software capable of analyzing a two-stroke engine carburetion.

Preview:

The program is capable to find out if the image is a spark plug or not.