AI Data Annotator
I worked on an AI data annotation project focused on training computer vision models using the Atlas Capture platform. The objective of the project was to improve the model’s ability to recognise human-object interactions in short video clips. My role involved segmenting video data into meaningful events and accurately labelling each segment based on predefined guidelines. For example, I identified when a subject initiated movement toward an object, when interaction began, and when the interaction ended or changed. Each segment required precise timing and correct classification to ensure the model could learn patterns effectively. I was responsible for reviewing multiple clips daily, ensuring consistency across annotations, and maintaining a high accuracy rate. I also conducted quality checks on previously labelled data, corrected inconsistencies, and incorporated feedback from quality reviewers to improve performance. The project required strong attention to detail, the ability to follow complex annotation rules, and efficient time management to meet daily targets. As a result of my contributions, the dataset quality improved, supporting the development of a more accurate and reliable AI model.