6 reactions
Scientists from the University of Washington and Princeton University have developed a compact camera using meta-lenses and optical computing, enabling object identification at the speed of light while significantly reducing power consumption.
The camera replaces traditional lenses with 50 layers of meta-lenses, functioning as an optical neural network that processes visual data 200 times faster than conventional computer vision systems, with comparable accuracy.
Researchers demonstrated that their nanophotonic neural network achieves 72.76% accuracy on CIFAR-10, surpassing AlexNet (72.64%), proving its potential for deep-learning-driven image recognition at high speeds.
The camera replaces traditional lenses with 50 layers of meta-lenses, functioning as an optical neural network that processes visual data 200 times faster than conventional computer vision systems, with comparable accuracy.
Researchers demonstrated that their nanophotonic neural network achieves 72.76% accuracy on CIFAR-10, surpassing AlexNet (72.64%), proving its potential for deep-learning-driven image recognition at high speeds.
| 24,000 mlx | |
| 14 | |
| 6 |
That’s groundbreaking! A compact, ultra-fast camera using meta-lenses and optical computing could revolutionize AI vision systems. Faster, more efficient, and even outperforming AlexNet, impressive! 🚀📸
