eoinedge/spark-plug-classification

Spark Plug Condition Classification

Labeled images of spark plugs for training a multi-class visual classifier. Includes four categories:

• normal
• carbon_fouled
• oil_fouled
• mechanical_damage

Optimized for use in Edge Impulse with the Transfer Learning (MobileNetV2) block.

• Input: Images (96x96)
• Task: Multi-class classification
• Use case: Visual diagnostics and condition monitoring

Model trained and demonstrated on Edge Impulse.

Looking for anomaly detection?
See the companion dataset: Spark Plug Anomaly Detection