Industrial Site Metal Object & Foreign Object Annotation Project
With over 2 years of experience in data labeling and annotation, I contributed to large-scale AI training projects focused on improving the recognition of metallic and foreign objects in industrial, construction, and mining environments Handled more than 50,000 images involving complex outdoor scenes, accurately identifying and labeling all metallic components such as vehicles, machinery arms, tools, and grates as metal_object. Misplaced or unwanted items, including drill rods, scrap metal, and industrial debris, were precisely annotated as both metal_object and foreign_object. Applied detailed polygon segmentation techniques to distinguish between true metallic surfaces and surrounding materials such as dirt, rocks, and shadows. Ensured consistency and quality through rigorous QA checks, maintaining an annotation accuracy rate above 98% across multiple batches of 1,000+ images each. Collaborated with project reviewers and team leads to refine labeling guidelines, resolve ambiguous