Unsupervised domain adaptive object detection for assembly quality inspection

A challenge to apply deep learning-based computer vision technologies for assembly quality inspection lies in the diverse approaches and restricted annotated training data. This paper describes a method overcoming by an unsupervised domain adaptive object detection model on synthetic images generated from CAD models unannotated captured cameras. On case study of pedal car front-wheel assembly, achieves promising results compared other state-of-the-art methods. Besides, is efficient implement production as it does not require manually