Human-in-the-Loop Enhancements for Artwork Object Detection

This study presents 'Human-in-Loop Object Detection' with a focus on zero-shot detection in Holocaust artwork using the CHORUS annotation framework. It evaluates models like OWLv2 and GroundingDINO, emphasizing challenges with generic labels and the importance of domain-specific curation. The CHORUS system combines AI with human input for refined analysis, featuring user-friendly annotation tools and data insights. Feedback from heritage experts highlights usability and ethical...