Overview of Synthetic Data Generation for Computer Vision in Industry

The introduction of Industrial Artificial Intelligence led us to adopt a variety of innovative technologies that add value to the product’s life cycle. One such breakthrough is computer vision-based quality inspection of manufactured products. The main challenge with implementing computer vision in industry is that of a shortage of dataset for training the neural networks. The main objective of this study is to provide an overview of the various methods available for synthetic data generation that can be utilized by industries. We emphasize on flexible production systems that produce unique products every time. The performance of an Object Detection model trained on real and synthetic datasets is used to validate one of the approaches.