Artificial Intelligence for Electrode Quality Control in Battery Manufacturing: Evaluating Machine Learning, Deep Learning, and Transfer Learning Models for Tin, Zinc, and Titanium Electrodes

This dataset consists of optical images of screen-printed electrodes collected for surface quality assessment in a laboratory manufacturing environment. The images are organized into three top-level folders, each corresponding to a different electrode material: Tin, Zinc, and Titanium.

Within each material folder, images are further grouped by surface quality class. For Tin electrodes, the classes are Tin_good, Tin_medium, and Tin_bad. For Zinc electrodes, the classes are Zinc_good and Zinc_bad, and for Titanium electrodes, the classes are Titanium_good and Titanium_bad. Each image represents an individual printed electrode surface.

All images were acquired using a consistent optical imaging setup under controlled lighting and magnification conditions following the screen-printing process. The dataset captures natural variations in print quality arising from process variability, including differences in surface homogeneity, edge definition, and defect presence. The dataset contains no synthetic or augmented images and reflects real inspection conditions encountered during electrode fabrication.

This dataset is intended to support research in machine learning, computer vision, and automated visual inspection for printed electronics and related manufacturing applications.