Meetup Computer Vision How to scale training Data

It s better to have a standard algorithm on a lot of good data than a stateoftheart algorithm on little Thus data labelling has become, even if very painful, an unavoidable step in the modeling process. However, scaled annotation requires a combination of intuitive interfaces and machine learning (for preannotation for example). Moreover, scaling without compromising data quality requires transparency throughout the labeling process to facilitate quality monitoring and collaboration internally a