Training perception models requires large and diverse datasets. Assembling these datasets can be costly, time-consuming, and even dangerous for certain corner cases. By leveraging the synthetic data generation capabilities of Isaac Sim, developers can bootstrap the training task. This video highlights Isaac Sim’s support for domain randomization which allows for random changes in texture, colors, lighting, and placement. The video also showcase the different types of data including bounding boxes, depth, and segmentation that can be generated from the simulation environment. https://developer.nvidia.com/isaac-sim