Synthetic Data-Trained Algorithm Sets a New Standard for Object Pose Estimation
1,922 views · Published 4 December 2018 · 1:12 · Indexed 7 October 2026
Channel: NVIDIA Developer · 2018 · Science & Technology
At NeurIPS 2018, NVIDIA researchers showcase a demo of how, for the first time, an algorithm trained only on synthetic data is able to beat a state-of-the-art network trained on real images for object pose estimation on several objects of a standard benchmark. Learn more: https://news.developer.nvidia.com/nvidia-research-puts-smart-picking-within-grasp/ See full demonstration here: https://www.youtube.com/watch?v=yVGViBqWtBI
More from this channel
-
3:29
Share Your Science: Quantum Transport Simulations on Hybrid Supercomputers
-
3:01
Share Your Science: Detecting Tumor Cells in Blood Samples
-
2:36
Share Your Science: Magnetic Nanotechnology in Biomedicine
-
2:19
Share Your Science: Accelerating Oil Discovery with GPUs
-
1:56
Share Your Science: Next-Generation Computational Fluid Dynamics Technology
-
1:30
Share Your Science: Real-Time Simulation and Prognosis of Smoke Propagation
-
5:42
3 Steps to More Science Guide – Step 3: Optimize
-
2:30
Share Your Science: Deep Learning in Medical Imaging