| dc.description.abstract | The data annotation process is a critical step in the development of machine learning (ML) models, as it entails labeling data to help supervised learning. This study investigates the impact of data annotation process requirements on the performance
of ML models. Employing an experimental approach, the study compares the performance of ML models using different annotated datasets and various process requirements. Performance metrics, including average precision, precision, recall, and
F1 score, are used to compare the outcomes. The study reveals that the requirements imposed on the data annotation process have a substantial influence on the performance criteria of ML models. These findings shed light on the crucial role that the data labeling process plays in the creation of ML models, providing valuable insights for both academic researchers and industry professionals. keyword-Data annotation, performance metrics, precision, recall, true positive, False negative, false positive, Intersection over Union, mean average precision, experiment methods. | en |
