Spark Plug Detection and Counting
Identify and count automotive parts on trays
Related Products
Powered by In-Sight ViDi Deep Learning-Based Vision Software
Graphical programming environment for deep learning-based industrial image analysis
Most vehicle manufacturers now rely on a mix of suppliers for original equipment manufacturer (OEM) spark plugs. Once a spark plug brand is factory approved for a given automobile model, it can be used interchangeably with others. One brand of spark plug may be installed in the automobile one day, another the next. At the same time, multiple spark plug types for various automobile models are constantly being delivered and need to be identified and tracked.
Spark plugs arrive in wide trays, which makes it hard to accurately detect and count the parts. Wide field of view lenses show spark plugs at the edge at different angles than the ones nearer the center. The trays themselves may also be of different colors and textures.
Conventional machine vision has trouble automating spark plug counting since the parts appear at various angles to the camera, making them hard to identify and count reliably with rule-based vision technology. This can introduce production line issues due to parts shortage.
Cognex Deep Learning’s part location tool trains on sample images of representative trays of the different spark plug brands and types, and quickly learns to identify and count various spark plugs from a variety of angles, against varying color backgrounds. Once the spark plus detection and counting process is complete, they can then be placed in inventory and be ready to be installed into the cylinder heads of the appropriate vehicle.
