Image Quality Audit Tool

Large batches of cropped collectible images needed to be reviewed for issues such as incorrect orientation, flipped images, failed crops, or unreadable results. Manual review alone could be slow and inconsistent, especially when working with high-volume image sets.

Create a workflow that could help prioritize manual review by surfacing images most likely to need attention.

Contributed to the design and development of an audit approach that combined image-processing techniques, OCR, feature matching, and machine-learning-assisted flagging to identify crops that may require review.

The audit workflow analyzed cropped images with multiple checks instead of relying on a single method. OCR helped detect readable text and possible orientation issues. Feature matching helped compare expected visual structure. Image-processing and machine-learning-assisted checks helped flag suspicious crops for manual review.

The tool supported the quality-control process by helping surface images most likely to contain crop, orientation, or visual-quality issues. This made review work easier to prioritize and helped improve consistency in the image audit workflow.

This project strengthened my understanding of image-processing workflows, quality-control automation, and how to combine multiple technical approaches to solve practical production problems.